Google Ads Starter FREE Course · Day 7 of 7 · Article 2 of 2
Launch, the first 30 days, and when to stop DIY-ing
Seven days, in the order a first account actually gets built. Every lesson is a full reference article with step-by-step blocks you can follow with Google Ads open in another tab.
This day's reading:
- Google Ads Learning Phase: The First 30 Days
- Google Ads Optimization: Which Lever To Pull (you are here)
All seven days
- Is Google Ads worth it for you? The maths
- How the auction works, and setting up without the traps 2 articles
- Measurement before money 3 articles
- Keywords and intent 3 articles
- Structure, budgets and bidding at launch 3 articles
- Ads and landing pages 3 articles
- Launch, the first 30 days, and when to stop DIY-ing (you are here) 2 articles
Prefer it in your inbox? Get the 7-day course by email — one lesson a day, no pitch until day seven.
When a campaign is already live and spending: which lever fixes which problem, how much movement to expect in impression share, click-through rate and cost, and how long to wait before judging it. Diagnose the constraint first.
Last updated: November 18, 2026
In This Guide
- Executive Summary
- Benchmarks & Numbers at a Glance
- The Metrics That Pick the Lever
- How the Auction Decides
- Diagnosing a Campaign in 30 Minutes
- The Complete Lever Map
- Budget Levers
- Bidding Levers
- Quality Levers
- Keyword & Match Type Levers
- Creative Levers & CTR
- Audience & First-Party Data Levers
- Performance Max Levers
- Demand Gen, Video & Display
- Shopping & Merchant Center Feed Levers
- Landing Page & Conversion Levers
- Sequencing & What Is Reasonably Possible
- What Changed Recently
- References
1. Executive Summary: Five Principles for a Failing Campaign
This article is a working reference for an advertiser or agency managing a live Google Ads account that is already spending money and not hitting its targets. It is not a launch guide. Every section assumes the campaigns are running, the pixels are firing, and the problem is performance, not setup. The five principles below set the frame for everything that follows. Each names a lever, states a realistic size of movement, and says what that lever cannot do.
- Principle 1 — Diagnose the constraint before you touch the budget. Search lost impression share (budget) and Search lost impression share (rank) are reported at campaign level under Competitive metrics. If lost IS (rank) is above 20% and lost IS (budget) is below 5%, adding money does nothing; the campaign is losing auctions, not running out of cash. Fix Ad Rank first. If lost IS (budget) is above 15% and the campaign is meeting its CPA or ROAS target, a budget increase of 20–30% can recover 10–18 impression share points within 3–5 days — but only if the numerator (impressions received) rises while the eligible pool stays roughly constant. Spending more into a poor-quality auction buys impressions you should not want.
- Principle 2 — Bidding strategy changes are the fastest lever with the worst failure mode. Switching from manual CPC to Target CPA or adjusting a Target CPA by more than 20% in one move can send the campaign into a learning period of up to two weeks, during which daily CPA volatility doubles and impression share moves unpredictably. The correct move size is 10–20% at a time, evaluated after at least 14 days and one full conversion-delay cycle. A campaign with fewer than 30 conversions per month is not ready for Target CPA; use Maximize Conversions without a target first and build volume.
- Principle 3 — Quality levers are slow but cheap. Moving a keyword’s Ad relevance rating from Below average to Average typically takes 7–21 days to register in the reported Quality Score after ad edits, because the diagnostic uses a 90-day rolling comparison. The practical payoff is a reduction in the bid required to hold a given position — a WordStream observational analysis found that Quality Score 3 keywords paid roughly 150% more than the expected CPC while Quality Score 7–8 keywords paid roughly 20–30% less, though these are correlational figures, not a Google-published causal formula. The ceiling is real: a keyword whose query mix is too broad for any single ad to match well will not reach Above average relevance without structural changes to the ad group.
- Principle 4 — Creative levers move CTR, not conversion rate; confusing them wastes months. Improving Ad Strength from Poor to Excellent is associated with 15% more clicks and conversions on average according to Google’s own aggregate claim,[1] but that figure combines clicks and conversions and reflects campaigns that also improved asset coverage, not just the rating. A realistic planning range for a focused creative refresh on an already-active campaign is 5–20% relative CTR improvement, readable after 2–4 weeks and several thousand impressions per variant. CTR improvement that is not paired with a matching landing-page offer does not move conversion rate and therefore does not move CPA.
- Principle 5 — Some campaigns are at their ceiling and no lever will help without a structural change. When Search impression share exceeds 70%, Search lost IS (rank) is below 10%, Search lost IS (budget) is below 5%, and conversion rate has been flat for 60 days, the account has likely saturated available demand at acceptable economics. No bid adjustment, ad rewrite, or negative-keyword pass will materially improve it. The only paths forward are: a wider geography, a broader match type paired with better conversion data, a better offer or price point, a landing-page redesign, or a new campaign type to reach users earlier in the funnel. This article names those ceilings explicitly in every section rather than offering another optimization that cannot clear them.
2. Benchmarks and Numbers at a Glance
7.52%
Average Google Ads conversion rate, all campaigns
5.42USD
Average Google Ads CPC, all campaigns
30conv.
Practical minimum monthly conversions for Target CPA
7-14days
Typical Smart Bidding learning period at adequate volume
The table below is the single reference for every benchmark cited in this article. Country of origin is stated where known. Figures from vendor-published benchmarks are labeled as such; they are useful for orientation but should not be treated as Google-published causal targets. Where a figure is used in a worked example, the row number in this table is cited at that point.
| Metric | Typical range or threshold | Applies when | Source |
|---|---|---|---|
| Average Google Ads CPC, all campaigns | $5.26–$5.42 | All campaign types; country not stated; source is Australian publisher; treat as directional for US | [2] |
| Average Google Ads conversion rate, all campaigns | 7.52% | All campaign types; country not stated; vendor claim from study of 16,000+ campaigns | [2] |
| Search CTR by industry (US-oriented vendor benchmarks) | 4.8%–10.0% | Range across tracked industries; Automotive 8.8%, Legal 4.8%, B2B Services 5.1%, Education 6.4%, Apparel 6.5%; country not stated; vendor claim | [3] |
| Search conversion rate by industry | 1.6%–10.2% | Apparel 1.6%, Automotive 5.7%, Legal 7.0%, Education 7.1%, B2B Services 4.9%; country not stated; vendor claim | [3] |
| Search CPA by industry | $33.50–$116.00 | Automotive $33.50, Education $72.70, Legal $86.00, B2B Services $116.00; country not stated; vendor claim | [3] |
| Display CTR, all industries | 0.39%–0.60% | Display campaigns; country not stated; Real Estate outlier at 1.08%; vendor claim | [3] |
| Display CPC, all industries | $0.44–$0.86 | Display campaigns; country not stated; vendor claim | [3] |
| Shopping conversion rate | 1.3%–3.3% | Shopping campaigns; country not stated; vendor claim | [3] |
| Smart Bidding learning period — typical duration | 7–14 days | Search campaigns with 50+ monthly conversions; Google’s Display guidance cites up to 4 weeks at under 30 conversions per month; vendor claim derived from Google documentation | [4] |
| Smart Bidding learning period — low volume | Up to 28 days | Fewer than 30 conversions per month; Google Display documentation; more volatility expected throughout | [4] |
| Practical minimum conversions for Target CPA stability | ~30 per month | Per campaign; Google recommends higher volumes for better automation; below 30 expect high CPA variance | [4] |
| Safe Target CPA / Target ROAS change size | 10–20% per move | Search Smart Bidding; larger changes increase learning period risk; Google published guidance | [4] |
| RSA Ad Strength Poor→Excellent: associated lift | ~15% more clicks and conversions on average | Google aggregate claim; combines clicks and conversions; not an isolated CTR figure; vendor claim (Google) | [1] |
| Mobile page load delay — bounce rate impact | +90% bounce rate at 1s→5s load time | Mobile sites; relative increase in bounce probability; Google Mobile Speed research; not a universal conversion-rate formula | [5] |
| Mobile page load delay — conversion rate impact | Up to –20% mobile conversions per 1-second delay | Mobile landing pages; directional benchmark; Google published claim; actual result varies by page type and baseline | [5] |
| Quality Score vs. CPC relationship (observational) | QS 1 ≈ +400% CPC; QS 10 ≈ –50% CPC | Search keywords; WordStream observational analysis; correlational, not a Google-published causal formula; treat as directional; source is dated | [5] |
3. The Share and Performance Metrics That Tell You Which Lever to Pull
Every optimization decision in this article starts with reading the right metric correctly. The table below defines each key metric, states exactly where it is reported in the Google Ads interface, names which campaign types surface it, and says what a bad number tells you to do next. Where a metric does not exist for a campaign type, the proxy to use instead is stated explicitly. Reading a Search-only metric and applying the conclusion to a Performance Max campaign is one of the most common diagnostic errors in live accounts.
| Metric | What it actually measures | Where it is reported | Campaign types that report it | What a bad number tells you to do |
|---|---|---|---|---|
| Search impression share | Impressions received ÷ estimated eligible Search impressions[6] | Campaigns → Columns → Competitive metrics → Search impr. share | Search only | If below 50%, check lost IS (budget) and lost IS (rank) to find the binding constraint before any other move |
| Search absolute top impression share | Impressions in position 1 ÷ estimated eligible absolute-top opportunities[6] | Campaigns → Columns → Competitive metrics → Search abs. top IS | Search only | If below 20% and conversions are position-sensitive, investigate rank loss at the absolute-top level specifically |
| Search top impression share | Top-of-page impressions ÷ estimated eligible top-of-page opportunities[6] | Campaigns → Columns → Competitive metrics → Search top IS | Search only | If Search IS is high but Search top IS is low, the campaign shows often but rarely above organic results; bid or quality is limiting position |
| Search lost impression share (budget) | Estimated eligible impressions lost because budget ran out[6] | Campaigns → Columns → Competitive metrics → Search lost IS (budget) | Search only | Above 15% with acceptable CPA: test a budget increase. Above 15% with poor CPA: fix conversion economics first, then budget |
| Search lost impression share (rank) | Estimated eligible impressions lost because Ad Rank was insufficient[6] | Campaigns → Columns → Competitive metrics → Search lost IS (rank) | Search only | Above 20%: check Quality Score components; if Ad relevance is Below average on keywords with over $100/month spend, restructure those ad groups before raising bids |
| Click share | Clicks received ÷ estimated maximum achievable clicks[6] | Campaigns → Columns → Competitive metrics → Click share | Search, Performance Max, Shopping, Hotel | Low click share with high impression share means competitors earn more clicks from the same auctions; audit offer, assets and ad copy |
| Display impression share | Display impressions received ÷ estimated eligible Display impressions[6] | Campaigns → Columns → Competitive metrics → Display impr. share | Display only | Low Display IS with acceptable ROAS: test budget or bid increase. Low IS with poor ROAS: fix creative and audience first |
| Click-through rate (CTR) | Clicks ÷ impressions for the reporting period | All campaign views → Columns → CTR (default column) | All campaign types | Falling CTR at stable position signals ad relevance or offer mismatch; compare by device and query type before editing copy |
| Conversion rate | Conversions ÷ clicks for the reporting period | All campaign views → Columns → Conv. rate | All campaign types (requires conversion tracking) | Below industry benchmark with strong CTR: the problem is on the landing page or in the offer, not the ad; pull landing-page lever before creative lever |
| Cost per conversion (CPA) | Cost ÷ conversions for the reporting period | All campaign views → Columns → Cost / conv. | All campaign types (requires conversion tracking) | Rising CPA with stable CTR and conversion rate means CPCs are rising; check auction competition, bid strategy targets and Search lost IS (rank) |
| Performance Max — no Search IS | Not reported; PMax IS is calculated across Search + Shopping, not as a standalone Search metric[6] | Use click share as the competitive proxy for PMax | Performance Max only (use click share instead) | Low click share in PMax: improve asset coverage, audience signals and feed quality; there is no rank/budget loss split available |
| Demand Gen — no impression share or click share | Neither Search IS nor click share is available for Demand Gen[6] | Use reach, frequency and viewability from Campaign reporting | Demand Gen only | Low reach with adequate budget: broaden audience or creative; high frequency with low CTR: creative fatigue; rotate assets |
Failure mode: The single most common mistake is reading a campaign-level impression share that has fallen and immediately raising the budget — without first checking whether lost IS (rank) is the larger number. When rank is the constraint, budget increases buy more auctions the ad loses, raising spend and CPA simultaneously while impression share barely moves.
Worked example (illustrative)
Reading the share metrics before touching the budget
- Setup: A Dallas HVAC company spending $6,500 a month on Search, running three campaigns targeting residential repair, maintenance contracts and new installation.
- Numbers: The installation campaign shows Search IS 41%, Search lost IS (budget) 4%, Search lost IS (rank) 55%. The repair campaign shows Search IS 68%, Search lost IS (budget) 19%, Search lost IS (rank) 13%. The account manager’s first instinct is to raise both budgets. At the account’s average CPC of $4.80[2], recovering 10 impression-share points in the installation campaign via budget alone would cost roughly $480 extra per day — but with rank loss at 55%, those auctions would mostly be lost anyway, adding spend without adding impressions. The correct move is to fix rank on the installation campaign first (restructure ad groups, improve Ad relevance) and raise budget only on the repair campaign where rank loss is 13% and budget loss is 19%.
- Decision: Budget on the repair campaign raised from $80/day to $96/day (+20%). Installation campaign budget held; ad groups restructured to tighten keyword-to-headline alignment over the next 14 days.
- Why: When Search lost IS (rank) dominates, the eligible denominator is not the problem — the ad’s competitiveness in those auctions is; more budget enters the same losing auctions and raises spend without raising impressions received.
4. The Auction: What Actually Decides Whether You Show and Where
Every impression, every position, and every asset that accompanies an ad is determined by Ad Rank — not by budget alone, not by keyword match type alone, and not by Quality Score alone. Understanding the components the reader can control, the ones they can only influence, and the ones they cannot touch at all is the prerequisite for every lever decision in sections 7 through 16.
The six Ad Rank components
Google describes Ad Rank as a function of six inputs evaluated at every individual auction.[7] There is no published numeric formula. The practical model is:
Ad Rank = f(bid, auction-time ad quality, Ad Rank thresholds, search context, expected impact of assets and formats, auction competitiveness)
Each component works as follows:
- Bid: The bid signal the system uses — either a manually set maximum CPC or the auction-time bid generated by a Smart Bidding strategy. A higher bid raises Ad Rank, but cannot overcome a large quality deficit. With Smart Bidding, the advertiser sets the strategy and targets; Google determines the per-auction bid.
- Auction-time ad quality: Google estimates expected click-through rate, ad relevance and landing page experience for the specific user, query, device, location and time — not from the reported Quality Score. The 1–10 Quality Score is a diagnostic that updates on a 90-day rolling window;[5] it is not what enters the auction. Auction-time quality can improve before the visible score moves.
- Ad Rank thresholds: Every position in every auction has a minimum Ad Rank that must be cleared. Thresholds are dynamic — they vary by position, topic, device, location and query.[4] A higher position has a higher threshold. Lower-quality ads can face higher thresholds. An ad may show at position 3 but fail the threshold for position 1 even with a higher bid.
- Search context: The specific query, its meaning, the user’s location, device, time of day and other signals all affect which ads are eligible and what the thresholds are. A commercial query has different thresholds from an informational one. A mobile query in Denver has different thresholds from a desktop query in Miami. The advertiser cannot control this component.
- Expected impact of assets and formats: Google estimates the incremental benefit of sitelinks, callouts, images, structured snippets and other assets. Assets can contribute to Ad Rank, which affects whether the ad is eligible to show with enhanced presentation.[8] Creating an asset does not guarantee it shows; Google determines eligibility at the auction level based on expected benefit, available space and context.
- Auction competitiveness: The other advertisers in the same auction, their bids, quality and expected asset impact all affect position and price. This is entirely outside the advertiser’s control.
What the reader directly controls, influences, and cannot touch
| Ad Rank input | Advertiser’s level of control | Primary levers available |
|---|---|---|
| Bid / bidding strategy | Direct | Bid strategy, target CPA/ROAS, manual CPC, bid adjustments |
| Ad text and assets | Direct | Headlines, descriptions, sitelinks, callouts, images |
| Keyword and query targeting | Direct (Search/Shopping) | Match types, negatives, search themes (PMax) |
| Landing page content | Direct | Copy, speed, mobile UX, offer clarity |
| Ad relevance component | Influence (via ad and structure) | Ad group structure, headline-keyword alignment |
| Landing page experience component | Influence (via page changes) | Relevance, speed, trust signals, navigation |
| Expected CTR component | Influence (via ad quality) | Offer, creative, negative keywords, historical performance |
| Ad Rank thresholds | None | Cannot be set or viewed; set by Google per auction |
| Search context signals | None | Cannot control user device, location intent or query meaning |
| Competitors’ bids and quality | None | Visible only via Auction Insights report |
Why a higher bid is not always the answer: A competitor with substantially stronger auction-time quality can hold position 1 at a lower CPC than an advertiser bidding more but showing Below average Ad relevance. The minimum-price-to-clear-the-threshold mechanic means the winning advertiser pays the amount needed to beat the next competitor’s Ad Rank divided by their own quality factor — so higher quality lowers the price needed to hold any given position.[7] This is why fixing a Below average Quality Score component is often worth more than a 20% bid increase.
Failure mode: An advertiser who reads rising Search lost IS (rank) and responds by raising bids without auditing Quality Score components pays more per click to hold the same position or worse, while the underlying quality deficit — which is also lifting the threshold they must clear — remains unaddressed.
Worked example (illustrative)
Quality deficit that a bid increase cannot fix
- Setup: A Chicago managed IT services company spending $4,200 a month on Search, with 6 keywords all set to phrase match, one ad group covering “IT support,” “managed IT,” “network monitoring,” “cybersecurity services,” “help desk outsourcing” and “cloud backup solutions.”
- Numbers: Average CPC $5.47[3], Search IS 38%, Search lost IS (rank) 51%, Search lost IS (budget) 11%. All 6 keywords show Ad relevance: Below average. The account manager raises the target CPA from $110 to $140 to give the system more room. Two weeks later, Search IS is 44%, but CPA has risen to $138 and lost IS (rank) is still 44%. The root problem is that one ad tries to serve “cybersecurity services” and “help desk outsourcing” with identical copy — Google’s auction-time quality estimate for each individual query is weak. The incremental bid headroom bought some auctions but did not close the quality gap, so the threshold the ad must clear for position 1 remains elevated.
- Decision: Ad group split into three tightly themed groups (network/support, security, cloud/backup), each with dedicated headlines matching the query theme. Target CPA returned to $110. Quality evaluation period: 21 days.
- Why: Ad Rank thresholds for higher positions are dynamic and can be elevated for lower-quality ads;[4] raising bids enters more auctions but does not reduce the threshold, whereas improving ad relevance lowers the Ad Rank required to clear each position.
5. Diagnosing a Campaign in Thirty Minutes: From Symptom to Lever
The order of operations matters as much as the individual levers. Pulling a creative lever on a campaign whose primary constraint is budget loss wastes the advertiser’s time. The thirty-minute protocol below sets the sequence: read the share metrics first, then segment by device and query type, then check conversion tracking integrity before touching any bid or creative setting. The diagnostic table that follows maps each common symptom to its first and second lever, and states explicitly what not to do.
Thirty-minute diagnostic sequence
- Minutes 1–5: At campaign level, add the Competitive metrics columns: Search impr. share, Search lost IS (budget), Search lost IS (rank), Click share. Record the numbers. The campaign with the largest lost IS (rank) number is the one with a quality or bid problem. The campaign with the largest lost IS (budget) is the one with a funding problem. Do not touch either until the remaining steps are complete.
- Minutes 6–10: Open the conversion tracking summary (Goals → Conversions → Summary). Confirm that at least one primary conversion action recorded a conversion in the last 7 days. If the last conversion recorded is more than 14 days ago and spend is ongoing, the entire CPA and conversion rate diagnosis is unreliable — fix tracking before reading any efficiency metric.
- Minutes 11–18: For the worst-performing campaign by CPA, go to Audiences, keywords, and content → Search keywords → add Quality Score columns (Quality Score, Exp. CTR, Ad relevance, Landing Page Exper.). Sort by cost descending. Identify any keyword spending more than $100 per month with a Below average component. That keyword is the first structural lever. Do not adjust bids on it until the ad relevance issue is diagnosed.
- Minutes 19–24: Run the Search terms report (Insights and reports → Search terms) for the last 30 days. Sort by cost descending and filter for zero conversions. Any term with more than $50 in spend and zero conversions in 30 days is a negative-keyword candidate — but verify conversion lag before adding it; B2B accounts can have conversion windows of 30–90 days.
- Minutes 25–30: Check the bid strategy status. A campaign showing “Learning” in the Status column is in a learning period triggered by a recent change. Do not make another bid, budget or structural change until the learning period ends. Record what the last change was and when it was made.
| Symptom | What it usually means | First lever | Second lever | What not to do |
|---|---|---|---|---|
| Low Search IS, lost IS (budget) above 15%, lost IS (rank) below 10% | Campaign is meeting the auction but running out of money | Raise daily budget by 20–30% if CPA is on target | Consolidate budget from low-priority campaigns into this one | Do not raise bids; bid is not the constraint |
| Low Search IS, lost IS (rank) above 20%, lost IS (budget) below 5% | Campaign is losing auctions on Ad Rank, not on money | Fix Below average Quality Score components on highest-spend keywords | Raise Target CPA by 10–15% if quality fix takes more than 2 weeks | Do not increase daily budget; it buys more of the same losing auctions |
| High Search IS (above 70%), poor conversion rate | Ad is showing and being clicked; the problem is on the landing page or in the offer | Run a landing-page audit: message match, page speed, mobile UX | Check conversion tracking for double-counting or broken tags | Do not rewrite the ad; CTR is not the problem |
| Falling CTR at stable impression share and position | Ad copy has become stale or a competitor has introduced a stronger offer | Add new headline and description variants to the RSA; test a differentiated offer or proof point | Run an Auction Insights comparison to identify new entrants | Do not raise bids; position is stable so CPC will rise without fixing the click rate |
| Rising CPC at stable impression volume | Auction competition has increased or bid strategy is raising bids to maintain position | Check Auction Insights for new competitors; review bid strategy target vs. actual CPA | Tighten match types or add negatives to exit low-quality, expensive auctions | Do not lower bids immediately; check whether the incremental conversions justify the higher CPC first |
| High click volume, zero or near-zero conversions | Conversion tracking is broken, or the landing page has a critical failure | Verify conversion tag firing in Google Tag Assistant; check Goal → Conversions → Diagnostics | Test the landing page manually on mobile; check for form errors, broken CTAs or redirect loops | Do not pause the campaign based on a CPA that is infinite because tracking is broken |
| Strong Search performance alongside a Performance Max campaign targeting the same queries | PMax may be absorbing branded or high-intent queries that Search should own | Add brand exclusions to the PMax campaign; check click share by campaign | Add exact-match brand keywords to Search with a separate budget | Do not pause Search to “let PMax run”; branded Search traffic converts at much lower CPA in most accounts |
| Impression share has plateaued at 65–75% despite budget and bid increases | The eligible impression pool may be exhausted at current targeting; campaign near demand ceiling | Check Search term report for untapped related intent; consider broader match type | Evaluate geographic expansion or new audience targeting | Do not keep raising bids; the denominator is not growing and marginal auctions at the ceiling are expensive |
| CPA rising while conversion volume is stable | CPCs are rising faster than conversion rate is improving | Check Search lost IS (rank) trend; audit whether Target CPA was recently lowered too aggressively | Review bid strategy learning status; if in learning, hold all changes for 14 days | Do not lower Target CPA further; that restricts bids and reduces impressions, compounding the problem |
| Campaign that has never converted despite 30+ days of spend | Conversion tracking is likely broken, or the offer and landing page are fundamentally mismatched to the query intent | Audit conversion tag in Tag Assistant; manually test the entire conversion path | If tracking is confirmed correct, run a landing-page message-match audit against the top 10 search terms by spend | Do not optimize bids on a campaign with no conversion signal; Smart Bidding with no conversions optimizes toward clicks, not results |
Failure mode: An advertiser who skips the conversion-tracking check in minutes 6–10 and proceeds directly to bid and creative changes risks spending weeks optimizing a campaign whose CPA is reported as infinite or zero because the tag is broken — every downstream decision made during that period is based on a metric that was never going to move.
Worked example (illustrative)
Thirty-minute triage that finds two separate constraints in one account
- Setup: A Portland residential solar installer spending $11,000 a month across two Search campaigns — Brand and Non-Brand — plus one Performance Max campaign with the same conversion goal (qualified lead form submission).
- Numbers: Non-Brand Search: IS 44%, lost IS (rank) 48%, lost IS (budget) 8%; average CPC $6.20; conversion rate 3.1%; CPA $200 against a $175 target. Brand Search: IS 71%, lost IS (rank) 9%, lost IS (budget) 20%; CPA $38. PMax: click share 31%; CPA $190. Conversion tracking check reveals that the PMax campaign is firing the same lead-form tag as Non-Brand Search, but the Non-Brand Search campaign tag last fired 11 days ago — the tag is broken on a recently redesigned landing page. The 3.1% conversion rate is stale data; actual current rate is unknown. The first action is not a bid change: it is restoring the Non-Brand Search conversion tag. At the Brand level, budget loss of 20% with IS 71% and CPA $38 means raising Brand budget from $15/day to $18/day (+20%) should recover roughly 7–10 IS points within 3–5 days at an incremental cost of approximately $3/day × 30 = $90/month for an estimated 2–3 additional brand leads per month at $38 CPA — well within target.
- Decision: Restore Non-Brand conversion tag (day 1). Raise Brand daily budget from $15 to $18. Hold all Non-Brand bid and creative changes for 14 days until conversion data is clean.
- Why: When Search lost IS (budget) is the dominant loss metric for Brand and the CPA is well below target, the numerator (impressions received) can be lifted by funding the gap; for Non-Brand, no bid or creative decision is valid while tracking is broken.
6. The Complete Lever Map: What Each Lever Moves, How Fast and How Much
This section is the operational core of the article. It lists every meaningful lever available in a live account, what each one actually moves in the auction, how long you must wait before the data is stable enough to read, and how much movement is realistic. It does not list what a lever can theoretically do — it lists what it reliably does for an account that is already spending money and not performing well enough.
Two tables follow. Table 6a answers the measurement questions: what moves, in which campaign types, how fast, and by how much. Table 6b answers the decision questions: why it works, when to pull it, and when not to. The tables carry the same 20 levers in the same order, numbered so you can read straight across from one to the other. Rows are ordered by time to read the result, fastest first, so the levers at the top of both tables are the ones that give you an answer this week.
What each lever moves, and how fast
| Lever | What it moves | Campaign types | Time to read the result | Realistic size of effect |
|---|---|---|---|---|
| 1. Fix conversion tracking | Conversion count accuracy; Smart Bidding signal quality | All | 1–3 days to confirm tag fires; 2–4 weeks for bidding to restabilize | Can restore 20–100% of apparently missing conversions; no upper bound on bidding improvement |
| 2. Add negative keywords | Search impression share (denominator shrinks); CTR; conversion rate; CPA | Search, Shopping, PMax (campaign-level) | 3–5 days for traffic shift to show in data | CTR lift 10–40% relative where query mix is poor; CPA improvement 5–25% where irrelevant spend is material |
| 3. Raise daily budget | Search lost IS (budget); impression volume; conversion volume | All | 3–7 days | IS (budget) loss falls proportionally; marginal CPA rises 5–30% above average as share approaches 80%+ |
| 4. Add or replace headlines and descriptions | CTR; Ad Strength; eligible asset combinations | Search (RSA), PMax, Demand Gen | 7–14 days for directional CTR signal; 3–4 weeks for stable read | Poor-to-Excellent Ad Strength associated with ~15% more clicks and conversions on average[1]; 0–25% relative CTR improvement depending on baseline |
| 5. Add asset types (sitelinks, callouts, promotions, images) | Ad Rank; ad space occupied; CTR; asset eligibility | Search, PMax, Demand Gen, Display | 7–14 days | No universal Google-published CTR percentage; directional lift through increased ad space and relevance signals; sitelinks typically raise CTR 10–20% relative where previously absent |
| 6. Change bid strategy | Conversion volume; CPA; ROAS; impression share; spend distribution | All (strategy options vary by type) | 7–14 days to exit learning; full read at 4 weeks | Moving from Maximize Clicks to Target CPA with ≥30 conversions/month: CPA typically improves 10–30% relative; volume may fall 10–20% initially |
| 7. Raise or lower Target CPA | Auction participation; impression share; conversion volume; average CPA | Search, PMax, Display, Demand Gen, Video, App | 7–14 days per 10–20% change increment | Each 10–20% target rise typically adds 5–15% more conversions at proportionally higher average CPA; each 10–20% cut reduces volume 5–20%[8] |
| 8. Raise or lower Target ROAS | Conversion value; spend; ROAS; impression share | Search, Shopping, PMax, Display, Demand Gen | 7–14 days per increment | Lowering target ROAS by 10–20% typically expands volume 5–15%; raising it restricts volume by a similar margin; actual ROAS change depends on query mix and margin spread |
| 9. Improve ad relevance (restructure ad groups) | Ad relevance component; Quality Score; Ad Rank; CPC at given position | Search | 7–14 days for Quality Score component to update after restructure | Moving Ad relevance from Below average to Average or Above average can reduce effective CPC 10–20% at the same position; there is no published causal formula |
| 10. Change keyword match types | Eligible impression pool (denominator); CTR; conversion rate; query mix | Search | 7–14 days for traffic mix to stabilize | Broad-to-exact tightening typically raises CTR 15–40% relative and raises conversion rate 5–20% relative; shrinks impression volume 20–60% depending on query diversity |
| 11. Add or remove keywords | Eligible impression pool; impression volume; click volume | Search | 7–14 days for new keywords to accumulate data | Incremental reach varies by query demand; new keywords rarely move account-level CPA meaningfully unless they carry significant volume |
| 12. Add audience signals or first-party data | Smart Bidding signal quality; conversion prediction accuracy; CPA or ROAS efficiency | All (signal role varies: hard targeting in Display/Demand Gen; signals only in PMax/Search Smart Bidding) | 14–21 days for bidding system to integrate new signals | Customer Match lists with ≥1,000 matched users: CPA improvement 5–15% relative is plausible; no published universal figure |
| 13. Widen or narrow geography | Eligible impression pool; impression volume; conversion rate; CPA by location | All | 14–21 days for geographic performance to stabilize | Removing low-converting states or DMAs can lift blended conversion rate 5–20% relative; widening geography expands volume but often raises CPA 10–25% if new areas convert less well |
| 14. Change ad schedule | Impression volume by hour and day; CPA by time segment; wasted spend in zero-conversion windows | Search, Display, Demand Gen, Video (not PMax dayparting) | 14–21 days to build a stable hourly conversion read | Eliminating a 20% spend share with zero conversions and reinvesting it in peak hours can reduce blended CPA 5–15%; effect depends on how concentrated conversion activity is |
| 15. Improve landing page experience | Landing page experience component; conversion rate; Quality Score; Ad Rank; CPA | All (conversion effect universal; Quality Score component is Search only) | Days to weeks for conversion-rate change; 1–4 weeks for Quality Score component to update | Fixing mobile load speed from 8s to under 3s: up to 20% conversion-rate improvement per Google’s directional benchmark[5]; full CRO projects: 10–50% relative lift where baseline page is poor |
| 16. Split or consolidate campaigns | Bidding signal pool; learning stability; impression share by segment; reporting clarity | All | 14–28 days post-change for learning to restabilize | Consolidating two campaigns with 15 conversions/month each into one with 30/month can reduce bidding volatility and improve CPA stability by 10–20% relative; splitting rarely improves performance unless segments genuinely have different economics |
| 17. Add promotions or price assets | CTR; ad space; competitive differentiation in the SERP | Search, PMax, Shopping | 14–21 days for asset performance to accumulate | No universal Google-published CTR lift; directional 5–15% relative CTR improvement where competitors lack visible promotions and discount is genuine |
| 18. Change the Merchant Center feed (titles, GTINs, images) | Shopping impression share; CTR; product eligibility; relevance matching | Shopping, PMax (retail) | 1–3 days for feed to process; 14–21 days for traffic data to stabilize after changes | Title restructure for a poorly named catalog: 10–30% relative CTR improvement plausible; GTIN completion can restore eligibility on surfaces requiring product IDs; no universal published figure[9] |
| 19. Improve the offer (price, guarantee, free trial, delivery) | Conversion rate; CTR; Quality Score components; CPA; ROAS | All (offer affects every surface) | 21–42 days for conversion-rate change to be statistically readable | A genuinely better offer (e.g., free returns vs. paid, same-day delivery vs. 5-day) can lift conversion rate 15–50% relative; the platform cannot manufacture this improvement |
| 20. Raise Search IS by raising bids (manual or via higher Target CPA) | Search impression share; Search top IS; Search lost IS (rank); CPC; CPA | Search | 3–7 days for IS to reflect bid change | Each 10% bid increase typically lifts Search IS 2–8 percentage points where rank is the constraint, but raises average CPC 5–12%; marginal CPA rises non-linearly above 70% IS[6] |
Figure 1
How long before each lever tells you anything
- 1. Fix conversion tracking1–3 days
- 2. Add negative keywords3–5 days
- 3. Raise daily budget3–7 days
- 20. Raise IS via higher bids3–7 days
- 4. Add/replace headlines and descriptions7–14 days
- 5. Add asset types7–14 days
- 6. Change bid strategy7–14 days
- 7. Raise or lower Target CPA7–14 days
- 8. Raise or lower Target ROAS7–14 days
- 9. Improve ad relevance7–14 days
- 10. Change match types7–14 days
- 11. Add or remove keywords7–14 days
- 12. Add audience signals / first-party data14–21 days
- 13. Widen or narrow geography14–21 days
- 14. Change ad schedule14–21 days
- 17. Add promotions or price assets14–21 days
- 18. Change Merchant Center feed14–21 days
- 15. Improve landing page experience7–28 days
- 16. Split or consolidate campaigns14–28 days
- 19. Improve the offer21–42 days
Bars are comparable because every one is measured in days (the top of each range drives the bar length). The size of the effect each lever produces is not shown and is not comparable across levers — see table 6a for effect ranges.
When to pull each lever, and when not to
Table 6b lists the same 20 levers in the same order. The Why it works column gives the auction mechanic or arithmetic that makes the lever move. The Use it when column is a testable account condition with a metric and threshold. The Do not use it when column names a specific disqualifying state — not a generic caution.
| Lever | Why it works | Use it when | Do not use it when |
|---|---|---|---|
| 1. Fix conversion tracking | Smart Bidding optimizes toward the conversion actions in the Conversions column; phantom or duplicated conversions feed it false signal | Conversion count differs by >10% from CRM or revenue data, or conversion tags fire on page load rather than on action | Tracking is verified accurate within the last 30 days — fixing it again risks introducing new errors; pull lever 2 or 3 first |
| 2. Add negative keywords | Removing irrelevant queries shrinks the eligible pool, raising CTR and concentrating spend on converting intent | Search terms report shows queries with zero conversions consuming >5% of monthly spend, or CTR is below 3% on a non-brand Search campaign | Search lost IS (rank) is already >30%; negatives will shrink the denominator further and make rank loss appear worse than it is |
| 3. Raise daily budget | Budget is a delivery ceiling; raising it allows the campaign to enter auctions it was previously exiting mid-day | Search lost IS (budget) >10% and current CPA or ROAS is at or better than target | Search lost IS (budget) is under 5%, or CPA is already above target — more spend compounds the efficiency problem |
| 4. Add/replace headlines and descriptions | More unique assets give Google more combinations to test; better offer language raises expected CTR at auction time | Ad Strength is Poor or Average, or fewer than 10 unique headlines exist in the RSA, or impression-weighted CTR is below industry benchmark | Ad Strength is already Excellent and CTR is at or above benchmark — creative is not the constraint; check bid strategy or landing page instead |
| 5. Add asset types | Assets expand ad space on the SERP and contribute to Ad Rank through expected format impact[8] | Sitelinks, callouts or structured snippets are absent, or the account has fewer than 4 sitelinks active on a Search campaign | Assets already exist and are eligible but impressions are low — asset addition will not fix a budget or rank constraint; fix lever 3 or 20 first |
| 6. Change bid strategy | Strategy determines what the auction-time bid optimizes toward; misalignment wastes spend on clicks that do not convert | Campaign has ≥30 conversions/month and is on Maximize Clicks or Manual CPC with no conversion objective | Fewer than 15 conversions/month — Smart Bidding will be highly unstable; fix tracking (lever 1) or consolidate campaigns (lever 16) first |
| 7. Raise or lower Target CPA | Target CPA sets the bid ceiling the system will accept per conversion; raising it allows higher bids and more auction participation | Actual CPA is consistently 15%+ below target (room to scale) or Search lost IS (rank) >20% with budget headroom | Conversion tracking is broken or fewer than 20 conversions in the last 30 days — the target has no reliable signal to optimize against |
| 8. Raise or lower Target ROAS | Target ROAS constrains bids to the conversion-value multiple specified; loosening it lets the system bid into lower-ROAS auctions to gain volume | Actual ROAS is consistently 20%+ above target and conversion volume is below business need | Conversion values are not set, are all equal, or are unreliable — ROAS optimization requires accurate value signals; use Target CPA instead |
| 9. Improve ad relevance | Tighter keyword-to-ad alignment raises Ad relevance component, reducing the bid required to achieve a given Ad Rank | Ad relevance reads Below average on keywords carrying >$100/month of spend in the Search keywords Quality Score view | Ad relevance reads Average or Above average across all significant keywords — the lever is already pulled; CTR problem is likely expected CTR, not relevance |
| 10. Change match types | Match type controls which queries enter the eligible pool; exact/phrase restricts the denominator to higher-intent auctions | Search terms report shows broad keywords matching irrelevant categories and negatives alone cannot clean the traffic | Broad match is paired with Target CPA/ROAS and ≥50 conversions/month — the bidding system is already filtering by intent; tightening match types will reduce its signal pool |
| 11. Add or remove keywords | Adding keywords expands the eligible auction pool; removing them concentrates budget on higher-performing queries | Search terms report shows recurring high-converting queries that have no matching keyword in the account | Campaign is already on broad match with Smart Bidding and ≥50 conversions/month — the system finds those queries anyway; duplicate keywords create routing confusion |
| 12. Add audience signals / first-party data | First-party lists give Smart Bidding stronger conversion-probability signals at auction time, reducing wasted spend on low-probability users | Customer Match list of ≥1,000 matched users is available and not yet uploaded, or remarketing audiences are absent from a campaign with >$5,000/month spend | List has fewer than 500 matched users — the signal is too thin to change bidding behavior meaningfully; build the list before uploading |
| 13. Widen or narrow geography | Restricting to converting geographies raises the proportion of spend in auctions where conversion probability is higher | Location performance report shows ≥2 states or metro areas with CPA >2× account average and >5% of monthly spend each | All geographies are within 20% of average CPA — location is not the constraint; cutting geography will shrink volume without improving economics |
| 14. Change ad schedule | Removing zero-conversion time windows prevents spend from being consumed in auctions the account cannot monetize | Hour-of-day segment shows ≥15% of spend concentrated in hours with zero conversions over a 30-day window | Fewer than 60 conversions in the last 30 days — the hourly sample is too small to distinguish zero-conversion hours from low-probability hours; a premature cut will remove valid auctions |
| 15. Improve landing page experience | A faster, more relevant, more usable page raises conversion rate and Landing page experience component, lowering effective CPA at unchanged CPC | Landing Page Exper. reads Below average on keywords with >$200/month spend, or mobile page load time exceeds 5 seconds (via PageSpeed Insights) | Landing page experience reads Above average and CRO testing has already been run — the conversion constraint is the offer (lever 19), not the page |
| 16. Split or consolidate campaigns | Pooling conversion volume into fewer campaigns gives Smart Bidding a larger, less volatile signal; splitting separates incompatible economics into independent budgets | Two or more campaigns share the same bid strategy but each has fewer than 20 conversions/month — consolidate; or brand and non-brand are in one campaign with different target CPAs | A consolidated campaign already has ≥50 conversions/month and stable CPA — splitting it will restart learning and reduce data per campaign below the stability threshold |
| 17. Add promotions or price assets | Promotion annotations occupy additional SERP space and visually differentiate the ad, increasing expected CTR for price-sensitive intent | Competitors in the auction have visible promotion badges and the account does not, or a genuine discount of ≥10% exists and is not surfaced in the ad | No genuine promotion exists — Google may disapprove fabricated promotions and policy violations risk the account, not just the asset |
| 18. Change Merchant Center feed | Accurate titles and GTINs expand the query surface that products are eligible to match; better images raise Shopping CTR | Merchant Center Needs attention tab shows >5% of active products disapproved, or Shopping CTR is below 1.5% with ≥500 impressions/month per product group | Feed has no warnings and Shopping CTR is at or above category benchmark — feed is not the constraint; check Shopping bids or campaign budget instead |
| 19. Improve the offer | A stronger offer raises conversion rate independent of ad or landing page quality; it is the only lever that can move a ceiling constrained by what the business sells | Conversion rate is below 2% on a purchase-intent Search campaign and landing page experience reads Average or Above average — the page is not the problem | Conversion rate is already at or above industry benchmark[1] — the constraint is traffic volume or margin, not conversion rate; pull lever 3 or 20 instead |
| 20. Raise IS via higher bids | A higher bid raises Ad Rank, allowing the ad to clear higher thresholds and win more auctions where rank was the constraint[7] | Search lost IS (rank) >20% and Search lost IS (budget) <5% — rank is the binding constraint, not budget | Search lost IS (budget) >10% — a bid increase will win more expensive auctions but the campaign will hit the budget ceiling faster; raise budget first (lever 3) |
The three levers most commonly pulled first that should not be: Advertisers routinely start with raising bids (lever 20), adding keywords (lever 11), and refreshing ad copy (lever 4) — in that order. All three are wrong as a first move in most underperforming accounts. Raising bids when Search lost IS (budget) is already above 10% burns faster through a budget that is already the binding constraint. Adding keywords when broad match and Smart Bidding are already running adds routing complexity without adding reach the system was not already finding. Refreshing ad copy when conversion tracking is broken teaches the bidding system nothing new. The correct first move is almost always lever 1 (verify tracking), then lever 2 (clean negative keywords), then lever 3 (address the actual share constraint). Pull levers 4, 11 and 20 only after the measurement and negative structure are clean.
Worked example (illustrative)
Reading the lever map to find the right first move
- Setup: A Dallas B2B software account spending $12,000/month on Search. Search lost IS (budget) is 4%, Search lost IS (rank) is 38%. The campaign is on Maximize Clicks with 11 conversions in the last 30 days. Ad relevance reads Below average on 6 of 9 keyword groups carrying $8,400 of the monthly spend.
- Numbers: With 11 conversions/month, Smart Bidding is below the 15-conversion floor for stability. The rank loss of 38% says the ad is losing roughly 38 of every 100 eligible auctions to Ad Rank, not budget. The B2B Services benchmark CPC is $5.47 and conversion rate is 4.9%[1]; at $12,000/month that is ~2,194 clicks, implying ~107 conversions if the benchmark applied — the gap to 11 actual conversions suggests a tracking problem, not a scale problem.
- Decision: Fix conversion tracking first (lever 1). Then add negatives from the Search terms report (lever 2). Do not raise bids (lever 20) or change bid strategy (lever 6) until tracking is verified and conversion volume is above 20/month.
- Why: Raising bids against a 38% rank loss with broken tracking feeds false signal to Smart Bidding; the auction mechanic means higher bids will raise CPC but the system cannot learn which of those more expensive clicks actually convert.
Managed for you
Everything on this page, handled every month
You are reading how to do it properly. Omologist does it, inside a written mandate, for 15% of your budget.
✓ Conversion tracking built and tested before any ad runs
✓ Search campaigns built to what your budget can actually feed
✓ Changes made inside a written mandate, not whenever
✓ A report on the 1st that says what was done and why
15% of budget plus GST · minimum A$330 incl GST · accounts under about A$10,000 a month
7. Budget Levers: Lost Impression Share (Budget) and What More Money Actually Buys
A budget increase is the fastest lever to execute and the easiest to misuse. This section separates the two questions it actually answers, explains the diminishing-return arithmetic as share rises, and tells you when to stop buying share.
Question one: is budget the constraint at all?
Open the campaign-level columns and add Search lost IS (budget) and Search lost IS (rank). These two figures tell you where the eligible impressions are going. If Search lost IS (budget) is below 5%, budget is not the binding constraint — adding money will not materially move impression share. If Search lost IS (rank) is above 20%, the campaign is losing auctions to Ad Rank, and more money enters those auctions but loses them at the same rate, buying more spend at the same CPA or worse.
The share arithmetic matters here. Search impression share is the ratio of impressions received to estimated eligible impressions. When you raise the budget, the numerator rises (more auctions entered and won), but the denominator does not change unless the campaign’s targeting, bids or quality also change. That means a pure budget increase that moves Search lost IS (budget) from 25% to 5% should lift Search impression share by roughly 20 percentage points — provided rank, quality and targeting are unchanged.[6]
Question two: is the marginal impression worth buying?
Budget increases do not produce uniform returns. The first tranche of additional spend recaptures the highest-quality missed auctions — the ones the campaign was eligible for and competitive in, but exited mid-day when the budget ran out. As share rises past 60–65%, the remaining missed auctions are increasingly those the campaign entered later in the day, in less-competitive time windows, or against lower-intent queries it was borderline eligible for. Marginal CPA rises because marginal conversion probability falls.
There is no published universal curve for this relationship. The arithmetic below assumes a campaign where average CPA at 60% Search IS equals the account target, and estimates what happens as share rises:
| Search IS reached | Approx. additional spend vs. 60% IS baseline | Estimated marginal CPA relative to target | Recommendation |
|---|---|---|---|
| 60% (baseline) | $0 | At target (1.0×) | Established baseline |
| 70% | +17% | ~1.1–1.2× target | Usually acceptable if target has room |
| 80% | +33% | ~1.2–1.5× target | Test carefully; monitor marginal CPA weekly |
| 90% | +50% | ~1.5–2.0× target | Stop unless target allows significant flexibility |
| 95%+ | +58%+ | Often 2.0× target or worse | Economically indefensible for most accounts |
These are illustrative ranges, not observed results. The relationship is highly account-specific: a branded Search campaign with strong conversion intent may sustain near-target CPA at 90% IS; a competitive non-brand campaign in a dense market may see marginal CPA deteriorate at 70%. The correct test is to calculate marginal CPA directly:
Marginal CPA = (New monthly spend − Baseline monthly spend) ÷ Incremental conversions in the same period.
If marginal CPA is below your target, continue raising budget. If it is above your target, stop — and investigate whether Ad Rank improvement (lever 20 paired with lever 9) or conversion-rate improvement (lever 15) would let you buy the same impression share more cheaply.
When a campaign is at its economic ceiling
Some campaigns cannot be profitably scaled further with budget alone. The honest signs of a budget ceiling are: Search lost IS (budget) is below 5% (budget is not the constraint), Search lost IS (rank) is above 25% (rank is the constraint), marginal CPA at the current share is already above target, and the offer or landing page has not changed. When all four of those conditions apply simultaneously, adding budget buys more expensive impressions in auctions the campaign is already losing. The correct response is not a budget increase — it is fixing rank (levers 9, 15, 5) or accepting that the economically recoverable share is lower than the theoretical maximum.
Failure mode: Raising daily budget when Search lost IS (rank) is the dominant constraint sends more money into auctions the campaign immediately loses, raising spend at the same or worse conversion rate and driving CPA above target within days.
Worked example (illustrative)
Calculating whether a budget increase is worth buying
- Setup: A Phoenix commercial cleaning account spending $9,000/month on Search. Search IS is 61%, Search lost IS (budget) is 22%, Search lost IS (rank) is 17%. CPA is $118 against a target of $140. The campaign is on Target CPA at $130.
- Numbers: At $9,000/month and a $5.42 average CPC[1], the campaign receives roughly 1,661 clicks. At a 7.52% conversion rate[1] that is approximately 125 conversions. Raising budget to eliminate the 22% IS (budget) loss implies adding roughly 22 percentage points of IS, moving from 61% to ~83%. The additional eligible impressions are (22/61) × 1,661 = approximately 599 extra clicks/month. At 7.52% conversion rate that is ~45 extra conversions. At the current $118 CPA, the incremental spend needed is approximately $5,310 (45 × $118). Actual marginal CPA = $5,310 ÷ 45 = $118, still below the $140 target. The budget increase is worth buying at this share level.
- Decision: Raise daily budget from $300 to $477 (the $9,000-to-$14,310 monthly equivalent). Set a 14-day review: if actual CPA on the incremental spend exceeds $140, hold at that budget level and address the rank loss (17%) with ad relevance and asset improvements before buying further share.
- Why: Budget is the binding constraint (22% IS budget loss vs. 17% rank loss); the marginal impression is profitable at this share level because the conversion rate is above average and the account has not yet reached the diminishing-return zone above 80% IS.
Worked example (illustrative)
Stopping the budget increase at the economic ceiling
- Setup: The same Phoenix commercial cleaning account, 90 days later. Budget has been raised and Search IS is now 83%. Search lost IS (budget) is 4%, Search lost IS (rank) is 13%. Monthly spend is $14,200. CPA over the last 30 days is $134 — still below the $140 target but rising.
- Numbers: To reach 90% IS from 83%, the remaining 7 percentage points of budget-recoverable impression share require approximately (7/83) × $14,200 = $1,197 more monthly spend. At the last 14-day marginal CPA of $158 (above the $140 target), those incremental conversions cost $158 each. Marginal CPA ($158) exceeds the target ($140) — the budget ceiling has been reached. The 13% rank loss represents the remaining gap and can only be closed by improving Ad Rank, not by adding more money.
- Decision: Hold budget at $14,200. Redirect effort to ad relevance restructuring (lever 9) and landing page improvements (lever 15) to reduce the 13% rank loss. Do not raise budget again until marginal CPA returns below $140.
- Why: Above 83% IS in this account the marginal impression is priced above the target CPA; the auction mechanic means competing for the final 10–17% of share requires bidding into progressively less-competitive or less-converting auctions where the account’s rank disadvantage is largest.
When to pull it, and when not to
Pull the budget lever when: Search lost IS (budget) is above 10% at campaign level; the current CPA or ROAS on the existing spend is at or better than the account target; and Search lost IS (rank) is lower than Search lost IS (budget), confirming budget is the larger constraint. Calculate the expected marginal CPA before raising budget and confirm it falls below the target. Raise in increments of 20–30% and re-read after 7 days before the next raise.
Do not pull the budget lever when: Search lost IS (budget) is below 5% — budget is inert as a lever and additional spend will find nowhere to go; or when Search lost IS (rank) exceeds Search lost IS (budget) — adding money buys entries into auctions the campaign is already losing, raising CPA without raising useful volume; or when current CPA is already above target — a budget increase scales the inefficiency and raises total wasted spend proportionally. In all three cases, fix the binding constraint first: rank, quality, or conversion economics.
8. Bidding Levers: Strategy, Targets and Lost Impression Share (Rank)
Bidding is the lever most advertisers reach for first and misuse most often. The strategy you choose determines what the auction optimizes toward; the target you set determines how aggressively it competes; and Search lost impression share (rank) tells you whether bidding is actually the binding constraint. These three decisions interact, and pulling any one of them badly resets the learning period and wastes the data the system has already accumulated.
Failure mode: Switching bid strategy or cutting a Target CPA by more than 20% in a single move triggers a learning period that can last two to four weeks, during which conversion volume drops, CPA swings unpredictably, and any performance read taken before the period ends will produce the wrong decision.[4]
The bid strategy table
| Bid strategy | What it optimizes | Effect on Search impression share | Use it when | Do not use it when | Failure mode |
|---|---|---|---|---|---|
| Maximize Clicks | Greatest click volume within budget | Can rise; bids compete broadly but not for position | Conversion tracking absent; fewer than 15 conversions/month; building initial data | Conversion data exists; CPA or ROAS matters; account is already spending efficiently | Traffic volume rises but CPA is uncontrolled; spend shifts to cheap, low-intent clicks |
| Maximize Conversions | Greatest conversion count within budget | Usually rises; system bids aggressively to find converters | 30+ conversions/month; budget is the primary ceiling; no efficiency target yet set | Fewer than 15 conversions/month; CPA target is firm; budget is already exhausted daily | Spends to full budget regardless of CPA; volume metric rises while unit economics collapse |
| Maximize Conversion Value | Greatest total conversion value within budget | Usually rises toward high-value auctions; low-value queries suppressed | 30+ conversions/month with reliable values; ecommerce or value-diverse leads | All conversions have equal value; values are estimated or inconsistent | Chases high-value single transactions; conversion count drops; ROAS uncontrolled without target |
| Target CPA | Conversions at a stated average CPA | Rises if target is raised; falls if target is cut | 30+ conversions/month; CPA target is known and economically defensible | Fewer than 30 conversions/month; target is set below current actual CPA | Target set too low: system under-bids, impression share collapses, conversions disappear |
| Target ROAS | Conversion value at a stated return on ad spend | Rises if target is lowered; falls if target is raised | 30+ conversions/month with reliable values; ROAS target matches business margin | Fewer than 30 conversions/month; conversion values are estimates or unmeasured | Target set too high: system restricts bids severely; volume and share fall while ROAS appears fine |
| Target Impression Share | A selected share threshold at a chosen position tier | Directly targeted; rises toward the set percentage | Brand defense; visibility campaigns where CPA is not the goal; 0 conversion data | Conversion efficiency matters; CPA or ROAS is the real objective | Bids aggressively to hit share target regardless of CPA; acquisition cost rises without limit |
| Manual CPC | Advertiser-set maximum CPC per keyword | Depends entirely on keyword-level bids set by the advertiser | Very low volume; granular bid control required; testing baseline; no Smart Bidding signals | Conversion data exists and volume is sufficient for automation; account is too large to manage manually | Under-bids winning auctions; over-bids losing ones; cannot adapt to real-time auction signals |
How large a target change is safe, and what resets learning
Google’s published guidance recommends changing Target CPA or Target ROAS by no more than roughly 10–20% in a single move, then waiting at least one full conversion-delay cycle—typically two weeks—before evaluating or moving again.[4] A larger change forces the system to re-learn the bid landscape from a materially different starting point. Moves that reset the learning period include: switching strategy entirely, changing the conversion action being optimized, moving the target by more than 20%, adding or removing a substantial share of traffic through new targeting, and large budget cuts or increases made simultaneously with a target change. During learning, expect daily CPA swings of 30–50% around the target; evaluate the two-week average, not day-by-day results.
The learning period runs approximately two weeks at 50 conversions per month and up to four weeks at fewer than 30 conversions per month.[4] Do not make a second change while the learning indicator is active; compounding changes makes it impossible to attribute the result to any single decision.
Reading Search lost impression share (rank) before touching the bid
Before raising a bid or switching strategy, read Search lost IS (rank) and Search lost IS (budget) at campaign level. If lost IS (rank) is above 20% and lost IS (budget) is below 5%, the constraint is Ad Rank—and the cheapest fix may be improving quality, not raising the bid. If lost IS (budget) is above 15% and the campaign is already profitable, the constraint is budget, not bidding. Raising a Target CPA when the real problem is budget wastes the learning period and delivers no impression-share recovery.
When to pull it, and when not to
Pull the bidding lever when: Search lost IS (rank) is above 20%; the campaign has at least 30 conversions per month; the current strategy’s target is set below the actual CPA achieved over the last 30 days; or you are switching from Maximize Clicks to Maximize Conversions because conversion tracking is now confirmed accurate. Change one thing at a time and confirm the target is at or above the 30-day actual CPA before tightening it.
Do not pull it when: Lost IS (budget) is above 15% and lost IS (rank) is below 5%—budget is the constraint and a strategy change will not recover share. Do not switch strategy or move targets when conversion volume is below 15 per month; Smart Bidding cannot learn from so little data and will produce volatile, unpredictable CPAs. Do not lower a Target CPA below the actual 30-day CPA in one move; pull the budget lever first if profitability is already acceptable.
Worked example (illustrative)
Target CPA reset recovering lost impression share (rank)
- Setup: A Dallas HVAC service account spending $14,000 a month on Search, running Target CPA at $95. The account target is $130 per booked call. Search lost IS (rank) is 31%, Search lost IS (budget) is 4%. The 30-day actual CPA is $118.
- Numbers: The target CPA of $95 is 19% below the actual CPA of $118, so the system is under-bidding most auctions to hit a target it cannot achieve. Raising the target by 15%—from $95 to $109—gives the system room to participate in auctions it is currently losing. At a 7.52% average conversion rate[1] and $5.26 average CPC[1], the 31% rank loss represents roughly 820 missed impressions per day at the campaign’s current CTR of 5.5%, equal to 45 missed clicks and 3.4 missed conversions daily, or approximately 102 conversions per month at an opportunity cost of $11,970 at the target CPA.
- Decision: Target CPA raised from $95 to $109 (a 15% move). No other changes made. Evaluate after 14 days of data.
- Why: The auction rejects bids that cannot clear the Ad Rank threshold for a given position; raising the CPA target gives Smart Bidding authority to submit higher bids in auctions it was previously losing on rank, without resetting the full learning cycle because the change is within the 20% safe-move boundary.[4]
9. Quality Levers: Expected CTR, Ad Relevance and Landing Page Experience
Quality Score is a 1–10 keyword-level diagnostic, not an auction input.[5] It tells you how your ad and landing page compare with other advertisers who showed for the same searches over the previous 90 days.[5] Improving it can reduce the CPC needed to hold a given Ad Rank position, but it is not a bidding target and it is not a primary KPI. The auction uses real-time quality signals, not the reported score.
Failure mode: Spending four weeks chasing Quality Score improvements on keywords that carry less than $200 of monthly spend each produces no measurable return; the same effort on two or three high-spend keywords with Below average components would recover meaningful CPC and impression share instead.
What each component is diagnostic of
Expected CTR measures whether your ad is likely to be clicked given the query. It is shaped by historical click performance for the keyword, ad-to-query alignment, offer clarity, and negative-keyword coverage. It moves slowly: because it is based on accumulated impression and click evidence over a rolling 90-day window, a copy edit alone will not move it in a week. Meaningful movement typically requires two to four weeks of impressions at sufficient volume—and “sufficient” generally means hundreds of impressions per week on the keyword in question.[5]
Ad relevance measures whether the ad matches the intent of the search. It is primarily a classification and alignment question. Because it reflects the relationship between keywords, ad group structure, and copy rather than accumulated click behavior, it can often be moved faster than expected CTR—sometimes within days of restructuring an ad group and rewriting the headline—once Google re-evaluates the new ad against new impressions.
Landing page experience measures whether the page delivers what the ad and query promised. Google evaluates relevance, transparency, navigability, mobile usability, and page load. It is usually the most expensive component to fix because it requires development, design, copy, and mobile work rather than a copy edit.[5]
Which component is cheapest to fix, and which is most expensive
Ad relevance is usually the cheapest fix: restructure an ad group by splitting unrelated keywords into their own group and rewrite the headline to match the query intent. This is a one-hour task. Expected CTR is a moderate effort: it requires creative testing at sufficient impression volume and may take several weeks to read. Landing page experience is usually the most expensive: meaningful improvement typically requires a developer, designer, and sometimes legal review, and the effect on the Quality Score status may take days after a material page change is crawled and impressions accumulate.[5]
CPC reduction from Quality Score improvement
An observational analysis by WordStream found that keywords at Quality Score 10 paid roughly 50% less than their expected CPC, while keywords at Quality Score 1 paid roughly 400% more.[5] This is not a Google-published causal formula—position, competition, bid, device, query, and assets are confounders—but the directional implication is real: improving a Below average component on a keyword spending $1,000 per month is worth pursuing even if the resulting share change is modest. Moving a keyword from Score 4 to Score 7 is not guaranteed to cut CPC by a fixed percentage; treat the relationship as directional and measure actual CPC before and after, holding targeting and bids constant.
There is no published figure for the typical CPC saving per Quality Score point; the arithmetic below assumes that correcting a Below average ad relevance status—by splitting the ad group and rewriting the headline—moves a keyword from Score 5 to Score 7, consistent with the WordStream directional relationship.
Account states in which chasing Quality Score is the wrong use of four weeks
Do not prioritize Quality Score work when: (1) Search lost IS (budget) is above 20% and the campaign is profitable—budget is the constraint, not quality, and fixing copy will not recover share; (2) the campaign has fewer than 30 conversions per month and Smart Bidding is unstable—fix conversion tracking and consolidate campaigns first; (3) the keywords with Below average components carry less than $100 each in monthly spend—the absolute CPC saving is too small to justify the time; or (4) the account is running Performance Max as the primary campaign type, because Quality Score is not reported for Performance Max and the diagnostic has no direct application.
How long Quality Score components take to move
Ad relevance can change within days of a copy rewrite once new impressions arrive. Expected CTR typically requires two to four weeks and meaningful impression volume before the status shifts. Landing page experience updates after Google recrawls the destination and observes new user behavior; allow at least one week after a material page change before reading the result, and confirm with actual conversion-rate data, which is a faster and more reliable signal than the Quality Score status itself.[5]
When to pull it, and when not to
Pull the quality lever when: Ad relevance reads Below average on keywords carrying more than $100 of monthly spend each; Expected CTR reads Below average and the ad group contains more than three unrelated intent themes; Landing page experience reads Below average and the bounce rate on the destination page is above 70% with average session duration below 20 seconds; or Search lost IS (rank) is above 20% and the bid is already at or above the industry benchmark CPC for the vertical.[1]
Do not pull it when: All three components read Average or Above average and lost IS (rank) is above 20%—the problem is bid level, not quality; when the keyword’s spend is below $100 per month and the absolute CPC saving is negligible; or when the campaign has fewer than 15 conversions per month and the highest-leverage fix is tracking and consolidation, not copy.
Worked example (illustrative)
Ad relevance fix recovering CPC on a high-spend keyword
- Setup: A Chicago commercial pest control account spending $6,500 per month on Search. The keyword “commercial pest control service” has Quality Score 4, Ad relevance Below average, Expected CTR Average, and Landing page experience Average. It spends $1,200 per month at a $7.40 CPC, generating 162 clicks and 11 conversions at a CPA of $109. The account’s target CPA is $120.
- Numbers: The keyword sits in an ad group with six other unrelated themes including “rodent exterminator” and “bed bug treatment.” The headline reads “Professional Pest Control Services” with no reference to commercial accounts. Splitting the keyword into its own ad group with the headline “Commercial Pest Control – Licensed & Insured” is expected to move Ad relevance from Below average to Average, consistent with the directional WordStream finding that a Quality Score improvement from 4 toward 6–7 reduces the effective CPC by approximately 15–20%. At the lower bound, $7.40 × 0.85 = $6.29 CPC. At $6.29 and the same 162 monthly clicks, monthly spend on this keyword falls to $1,019—a saving of $181 per month—while conversion volume holds if conversion rate is unchanged. That is $2,172 in annual recovered spend on one keyword.
- Decision: Created a new single-theme ad group for “commercial pest control” with two RSAs whose pinned headline 1 reads “Commercial Pest Control – Licensed & Insured.” Reviewed after 21 days once the new ad had accumulated 300+ impressions.
- Why: Ad relevance is primarily a classification signal comparing the keyword’s intent with the ad’s message; aligning them directly removes the mismatch that the 90-day diagnostic is flagging, and the improvement can appear within one to two weeks of new impression data.[5]
Not ready to change anything today?
We post the sharpest bits of these guides — what a change is worth, how long it takes to show up, and when to leave it alone. Short enough to read on your phone.
Follow on LinkedIn Follow on Instagram10. Keyword, Match Type and Search Term Levers
Match type is the lever that determines how wide a net the campaign casts, and therefore which eligible impression pool it competes in. Broadening match expands the eligible pool, which almost always lowers measured Search impression share in the short term even while increasing raw impressions—because the denominator of the share calculation grows faster than the numerator. This distinction is not optional knowledge; misreading a share drop after a match-type change as a performance problem has caused accounts to reverse a correct decision.
Failure mode: Adding broad-match keywords without confirmed Smart Bidding and without a negative-keyword framework in place causes the system to enter irrelevant auctions, spend clicks on queries that cannot convert, and compress the budget available to high-intent terms—all while measured impression share on the original terms falls, making the change look worse than it is.
Match type reference table
| Match type | What it now matches | Effect on impressions vs. measured share | Use it when | Do not use it when | Guardrail to apply |
|---|---|---|---|---|---|
| Broad | Queries related to the keyword by meaning, intent, or landing-page context; can match synonyms and implied intent | Impressions rise; measured Search IS usually falls as eligible pool expands | Smart Bidding active; 30+ conversions/month; conversion tracking confirmed; discovery of new intent desired | Manual CPC or Maximize Clicks in use; fewer than 30 conversions/month; negative list absent | Weekly search term review; negative keyword list active before launch; brand excluded if separate brand campaign exists |
| Phrase | Queries preserving the keyword’s core meaning; allows words before/after; reordering permitted if meaning unchanged (behavior expanded Feb 2021) | Moderate impression expansion; moderate IS denominator growth | Core intent must stay anchored; some query flexibility acceptable; moderate conversion volume | Intent must be very tightly controlled; exact coverage already sufficient | Bi-weekly search term review; check for close-variant drift into unrelated themes |
| Exact | Same intent including misspellings, plurals, abbreviations, function-word variants, and reordered wording (since Sep 2018) | Impressions constrained; IS denominator limited to close-intent pool; measured IS often highest | Brand terms; high-value high-volume terms; controlled experiments; known high-intent queries | Discovery is the goal; volume is insufficient to generate data from narrow targeting alone | Monitor search terms monthly; exact does not mean character-for-character—review variants served |
The share arithmetic when match type changes
Search impression share = impressions received ÷ estimated eligible impressions. When you broaden match, the eligible impression pool grows because more queries become eligible. If the new eligible impressions are in auctions where your Ad Rank is weak—as they often are, because the new queries are less aligned with the existing keywords, bids, and landing pages—the numerator grows slowly while the denominator grows quickly. The result is a measured share drop even though the campaign is showing more. Always check raw impressions, clicks, conversions, and CPA alongside share metrics after a match-type change. A share drop with rising conversions at an acceptable CPA is a successful broadening. A share drop with flat or falling conversions is a signal that quality or bid controls are needed.
Search term review: cadence and minimum volume before acting
Review the Search terms report weekly for campaigns spending more than $5,000 per month and bi-weekly for campaigns spending $1,000–$5,000 per month. For campaigns spending below $1,000 per month, monthly is acceptable because low volume makes individual query decisions unreliable. Before adding a negative, require at least 20 impressions with zero conversions on a query that is clearly irrelevant, or at least 50 clicks with zero conversions on a query that could be relevant. Do not exclude queries after one or two clicks—the signal is too thin. When adding a negative from the Search terms report, Google defaults to negative exact match; consider whether negative phrase better captures the irrelevant theme across variants.[7]
Keyword selection when multiple match types are eligible
Google does not apply a strict “most specific keyword wins” rule. When multiple keywords are eligible, it considers relevance, Ad Rank, expected performance, and Smart Bidding’s auction-time bid alongside match type.[7] Adding exact-match duplicates of broad-match keywords does not guarantee the exact keyword serves; it may simply create an internal competition that changes reporting without changing delivery. Add exact-match keywords only when there is a clear reason: a different landing page, a separate bid target, or a reporting separation that the account genuinely needs.
When to pull it, and when not to
Pull the match-type lever when: Search lost IS (rank) is above 20% and the keyword list is tight-match only with no broad discovery running; when the Search terms report shows high-converting queries not covered by existing keywords; or when you are switching to Smart Bidding and want to give the system a wider signal pool. Add broad match to the top five to ten highest-converting keywords first, with negatives in place, before expanding further.
Do not pull it when: Smart Bidding is in a learning period—broadening during learning compounds instability; when conversion volume is below 30 per month and the system cannot evaluate the new traffic quality; or when the budget is already exhausted on existing keywords, because broadening simply spreads a fixed budget thinner without adding impression-share recovery. Fix budget or bidding first, then broaden.
Worked example (illustrative)
Broad-match expansion: share drops, conversions rise
- Setup: An Atlanta IT managed services account spending $8,000 per month on Search, using phrase and exact match only, running Target CPA at $140. Search impression share is 54%, Search lost IS (rank) 28%, Search lost IS (budget) 6%. The account records 44 conversions per month at an actual CPA of $182—above the $140 target. The bid strategy is constraining auctions it should be winning.
- Numbers: The top five converting keywords are converted to broad match alongside the existing phrase and exact versions. A 60-term negative list is applied before launch covering competitor brand names, “free,” “DIY,” and informational prefixes. In the first 30 days after launch: impressions rise from 18,400 to 24,100 (a 31% increase); measured Search IS falls from 54% to 49% because the eligible pool expanded; conversions rise from 44 to 57 (a 30% increase); actual CPA moves from $182 to $140—on target—because Smart Bidding’s broader signal pool allowed it to find higher-converting auctions it was previously missing. At $5.26 average CPC[1], 57 conversions at $140 CPA = $7,980 of spend, within the $8,000 budget.
- Decision: Broad-match versions added to five highest-converting keywords; 60-term negative list applied at campaign level; search terms reviewed weekly for the first four weeks.
- Why: Broad match paired with Smart Bidding expands the eligible auction pool while letting auction-time bidding suppress low-probability impressions; the measured IS denominator grows faster than the numerator in the first two weeks, so the share metric falls even as the absolute conversion count rises—the correct read is on conversions and CPA, not share alone.[7]
11. Creative Levers: What Actually Lifts Click-Through Rate
Click-through rate is not primarily a creative variable. Position, query mix, device, brand recognition, and auction composition account for the majority of CTR variation across campaigns. The part of CTR that copy and assets can move is real but smaller than most advertiser attention to creative would suggest—and it can only be read after enough impressions accumulate to separate signal from noise.
Failure mode: Pausing a low-CTR RSA after 200 impressions and replacing it with a new one, then repeating the cycle, produces no learning and trains nothing; the account consumes creative budget and management time while the system never accumulates the impression volume needed to evaluate any asset.
What drives CTR and what copy can and cannot move
CTR = clicks ÷ impressions. The largest contributors to variance are: (1) ad position—absolute top ads receive materially higher CTRs than bottom-of-page ads, with Search absolute top impression share and Search top impression share the relevant metrics to check; (2) brand versus non-brand query mix—branded queries routinely generate CTRs two to four times higher than non-branded queries for the same account; (3) query intent specificity—transactional queries (“buy,” “hire,” “near me”) yield higher CTRs than informational queries for the same keywords; (4) device—mobile and desktop CTR differ by vertical and position. Only after controlling for these factors does ad copy contribute a distinguishable share of CTR movement.
There is no published Google decomposition of what percentage of CTR comes from position versus copy. The honest answer is that copy’s contribution is best measured by holding position, query mix, device, and landing page constant in a controlled experiment.
How many headlines and descriptions to run
An RSA accepts up to 15 headlines (30 characters each) and 4 descriptions (90 characters each).[10] Google recommends supplying all 15 headlines and all 4 descriptions, with genuine variation across assets—covering keyword relevance, benefits, proof points, objections, offers, and calls to action. The practical reason is combinatorial coverage: more distinct assets give the system more valid combinations to test across queries, devices, and positions. Moving from 6 headlines to 15 interchangeable rewrites of the same sentence does not expand combination quality; the variation must be substantive.[10]
Google reports that advertisers who improve Ad Strength from Poor to Excellent receive 15% more clicks and conversions on average.[1] This is an aggregate Google claim combining clicks and conversions, not an isolated CTR percentage, and it conflates creative quality with broader asset coverage improvements. Treat it as a directional ceiling for the full Poor-to-Excellent journey, not a per-metric guarantee.
Pinning: what it costs and when it is justified
Pinning forces an asset to a specific headline or description position, reducing the combinations available to the system. Google states pinning is generally not recommended and that it lowers Ad Strength.[1] If pinning is necessary—for legal disclaimers, pricing, or brand compliance—pin 2–3 distinct assets to the same position rather than a single rigid asset, so the system retains some combinatorial flexibility. There is no universally published CTR penalty for pinning; independent evidence is confounded because advertisers who pin often do so for compliance reasons that are correlated with other creative constraints. The honest expectation is that heavy single-asset pinning reduces combination quality and likely suppresses CTR and conversion rate below what full flexibility would produce, but the magnitude depends on how many positions are pinned and whether alternatives remain.[10]
Ad Strength: worth chasing or not
Ad Strength is a coverage diagnostic, not a performance predictor.[10] It reflects asset count, uniqueness, and combinatorial variety—not whether the message is compelling to the searcher. An account can have Excellent Ad Strength with 15 weak, repetitive headlines and poor conversion rates. The right use of Ad Strength is to identify gaps: missing asset types, duplicate wording, or pinning constraints. After fixing gaps, evaluate performance using impression-weighted CTR, conversion rate, and cost per conversion—not the Ad Strength rating itself. Chasing an Excellent rating by padding headlines with low-quality variations can dilute the system’s learning without any measurable CTR benefit.
Realistic CTR movements from a creative refresh
There is no single published figure for CTR lift from a creative refresh. The defensible planning range, based on the structure of the problem, is:
- 0–5% relative CTR improvement: likely when existing ads are already well-structured, assets are diverse, and the offer is competitive. This range is common and should be the base expectation.
- 5–15% relative improvement: plausible when assets are thin (fewer than 8 distinct headlines), copy is generic without offer specificity, or key intent themes are missing from the asset set.
- 15–30% relative improvement: possible when fixing a material mismatch—ads that do not mention the product category, lack a clear CTA, or have no differentiated offer against visible competitors. Not a dependable forecast; treat this as the upside scenario when the baseline is demonstrably weak.
These are reasoned estimates from auction mechanics, not published benchmarks. Where RESEARCH 10 supplies a figure, it is the Google aggregate claim of 15% clicks-and-conversions from Poor to Excellent Ad Strength.[1]
Minimum click volume before reading a creative test
For a directional CTR read, each test arm needs at least 300–500 clicks, covering the account’s normal weekly cycle to control for day-of-week effects. For a conversion decision, each arm needs at least 50 conversions—which at a 5% conversion rate requires 1,000 clicks per arm. At a $5.26 average CPC[1], that is $5,260 of spend per arm before the result is readable. An account spending $3,000 per month total cannot run a valid creative experiment in four weeks; use Google Experiments with a 50/50 traffic split and a pre-declared minimum detectable effect instead of pulling an ad based on early-week CTR data.
Assets: what actually earns more clicks
Sitelinks, callouts, structured snippets, and image assets expand the physical size of the Search result and communicate additional information before the click. Google does not publish a universal CTR percentage for any single asset type, and controlled evidence is thin because asset display depends on Ad Rank, format eligibility, device, and position.[8] The practical rule is: eligible assets that are not yet present are more likely to move CTR than refining copy that is already present. Add all eligible asset types before optimizing copy. Check asset eligibility by reviewing the Recommendations tab and the asset column in the Search keywords view; a campaign showing no sitelinks when sitelinks are eligible is leaving Ad Rank—and CTR—on the table.
When to pull it, and when not to
Pull the creative lever when: Ad Strength reads Poor or Average and the RSA has fewer than 10 unique headlines; Expected CTR reads Below average in the Quality Score column; impression-weighted CTR for a campaign is more than 15% below the industry benchmark for its vertical[1]; or eligible asset types (sitelinks, callouts, images) are absent from an ad group that has sufficient Ad Rank to display them. Run a Google Experiment with at least 300 clicks per arm before declaring a winner.
Do not pull it when: CTR is below benchmark primarily because Search absolute top impression share is low—the problem is position and bidding, not copy; when the account has fewer than 100 clicks per week per ad group, because there is insufficient volume to evaluate any creative change in a reasonable time frame; or when conversion rate is the constrained metric, because a CTR improvement that brings in lower-intent clicks will worsen conversion rate and raise CPA even as it makes the creative dashboard look better.
Worked example (illustrative)
Creative refresh economics: when CTR lift is and is not worth buying
- Setup: A Miami residential roofing account spending $9,000 per month on Search. Current Search CTR is 3.8%, against a Home Services benchmark of approximately 5%.[1] The RSA has 7 headlines and 2 descriptions; Ad Strength reads Average. CPC is $6.10, conversion rate is 6.5%, and the target CPA is $94. Monthly conversions: 96. Search absolute top IS is 61%—position is not the primary CTR constraint.
- Numbers: At $9,000 ÷ $6.10 CPC = 1,475 clicks per month. A 15% relative CTR improvement (from 3.8% to 4.4%) at the same impression volume would increase clicks to 1,710—adding 235 clicks per month. At 6.5% conversion rate, that is 15 additional conversions per month. At $6.10 CPC, those 235 extra clicks cost $1,434 in additional spend, raising total spend to $10,434. At 15 additional conversions, the incremental CPA is $1,434 ÷ 15 = $95.60—just above the $94 target. If CTR improvement is only 5% relative (3.8% to 3.99%), the increment is 78 clicks, 5 conversions, $476 of spend, and an incremental CPA of $95.20. In both scenarios the economics are marginal; the stronger leverage is in conversion rate, not CTR.
- Decision: RSA expanded to 15 headlines and 4 descriptions. Sitelinks added (were absent). Google Experiment created with 50/50 traffic split. Primary metric set as conversion rate, not CTR. Minimum 400 clicks per arm before read (approximately 27 days at current volume).
- Why: When the marginal CPA of incremental CTR-driven clicks is at or above the account target, the creative lever does not unlock profitable growth on its own—the landing page conversion rate is the binding constraint, and that must be addressed in parallel or first.[10]
12. Audience, First-Party Data and Signal Levers
Audience levers in Google Ads operate on a spectrum from hard targeting—where the audience boundary controls who can see the ad—to pure signal, where the boundary is advisory and the platform decides how far to extend it. Confusing the two is the most common mistake: an advertiser adds a Customer Match list, sees it labeled as “Signal” in Performance Max, and assumes the campaign is reaching only those users. It is not. Understanding which mode applies in which campaign type, and what the platform will and will not do with the data, determines whether first-party work is the right next dollar.
Targeting versus observation versus signal: the three modes
In targeting mode (Search bid adjustments, Display, Demand Gen with expansion off), the audience definition restricts or adjusts delivery. In observation mode (Search audience observation), the audience does not restrict delivery but enables bid adjustments and reporting for that segment. In signal mode (Performance Max, Demand Gen with optimized targeting on), the audience informs automation but the platform can and will serve beyond it. The table below maps modes by campaign type.
| Campaign type | Audience mode available | Hard delivery boundary? | Bid adjustment available? | Metric to judge it |
|---|---|---|---|---|
| Search | Observation (default), Targeting | Only in Targeting mode | Yes, in both modes | Conv. rate and CPA by audience segment |
| Display | Targeting, Observation | Yes, in Targeting mode | Yes | Conv. rate, CPA, view-through conv. |
| Performance Max | Signal only | No | No direct adjustment | Conversion quality; no audience IS metric |
| Demand Gen | Targeting, Optimized targeting | Yes, unless expansion on | No per-audience bid; campaign-level TCPA/TROAS | CPL, CPA, reach frequency |
| Video | Targeting, Observation | Yes, in Targeting mode | Limited | View rate, CPV, conv. if tracked |
| App | Signal only | No | No | CPI, in-app actions, ROAS |
Customer Match: requirements, realistic lift, and ceiling
Customer Match lets you upload hashed first-party identifiers—email, phone, name and address, or mobile device ID—which Google matches to signed-in Google accounts.[11] Match rates in US consumer accounts typically run 40–70%; B2B lists with work emails match at 20–45% because fewer users sign into Google products with a corporate address. There is no published Google figure for the conversion-rate lift from Customer Match alone; there is no published figure for this lever and the ranges below are reasoned from auction mechanics. Using a matched list in observation mode on Search allows a bid adjustment of −90% to +900%; a +30–50% adjustment on verified high-value customers typically moves impression share on those queries by 8–18 percentage points (numerator rises because bids are higher, denominator is unchanged) and can reduce CPA by 10–25% when the matched segment genuinely converts better. The result can be read after 14–21 days of sufficient matched-user auction volume—meaning the list must be generating at least 50 impressions per day before the signal is interpretable.
Customer Match requires the account to be in good standing and to meet Google’s eligibility requirements, including minimum spend history.[11] Lists are usable immediately after upload and processing (typically 24–48 hours), but a list of fewer than 1,000 matched users will not generate enough auction volume to move a measurable metric in less than 30 days for most Search campaigns.
Remarketing lists: minimum sizes, lag and realistic movement
Standard remarketing requires a minimum of 1,000 cookies (Display) or 1,000 active users in the past 30 days (Search RLSA). A fresh pixel on a site generating fewer than 3,000 monthly unique visitors will take 2–4 weeks to build a 1,000-user RLSA list. Once the list is usable, applying a +20–40% bid adjustment in observation mode on Search for cart-abandoners or high-intent page visitors typically lifts impression share on those users by 6–15 percentage points (numerator rises) and conversion rate by 15–35% relative, because the audience is self-selected high intent. Time to read: 14–21 days after list crosses the 1,000-user threshold.
Account states where first-party data work is not yet worth doing
First-party audience work produces no measurable return when: (1) monthly site visitors are below 2,000, so lists never reach usable size; (2) conversion tracking is broken or counting multiple actions per click, making CPA comparisons meaningless; (3) Search impression share is below 40% and Search lost IS (rank) is above 30%, meaning the binding constraint is Ad Rank, not audience selection—fix bidding and quality first; (4) the campaign has fewer than 30 conversions per month total, so audience-segment CPA is too noisy to interpret.
Failure mode: Applying a +70% bid adjustment to a Customer Match list before confirming that matched users actually convert at a higher rate than average buys expensive impressions with no efficiency gain—spend rises 10–25% with no corresponding CPA improvement.
When to pull it, and when not to
Pull it when: Search lost IS (rank) is below 15% (so rank is not the binding constraint); the account has at least 1,000 matched or cookied users in a list; conversion tracking is verified clean; and the campaign already converts at or better than the target CPA, meaning you are looking for efficiency or volume gains within a known-converting query set. Also pull it when first-party data can be used to suppress audiences—for example, adding current customers as a negative audience to a prospecting campaign to avoid paying for repeat conversions you would have received organically.
Do not pull it when: The list is below minimum size (the platform ignores sub-threshold lists silently—you will see no error); conversion tracking is miscounted; or Search lost IS (budget) is above 20%, meaning the campaign cannot fund the auctions it is already eligible for—fix budget first, because a bid adjustment on an audience segment inside a budget-limited campaign simply shifts spend from non-listed users to listed users without increasing total conversions.
Worked example (illustrative)
RLSA bid adjustment on a B2B lead-gen campaign
- Setup: A Chicago IT-staffing firm spending $6,200 per month on Search, targeting CPA of $95 for demo requests, currently averaging $103 CPA over 60 conversions in Q3 2026.
- Numbers: Site traffic: 4,800 unique visitors per month. RLSA list of pricing-page visitors: 1,340 users. Current campaign CTR: 5.1%[3], CPC: $5.47[3], conversion rate: 4.9%[3]. Matched-segment conversion rate from observation data: 9.2%—1.9× the campaign average. Applying a +40% bid adjustment raises effective max CPC from $5.47 to $7.66 on that segment. At 9.2% conversion rate that segment’s CPA is $7.66 ÷ 0.092 = $83, well inside the $95 target. Segment generates roughly 180 impressions per day; at the 5.1% CTR that is 9 clicks per day, adding roughly 10 conversions per month at $83 each—$830 of incremental spend for 10 incremental conversions.
- Decision: Set RLSA pricing-page list to observation mode; add +40% bid adjustment at ad group level. Evaluate after 21 days once the segment has accumulated at least 200 clicks.
- Why: The bid adjustment raises Ad Rank for the matched segment in proportion to the bid multiplier, increasing impression share numerator without changing the eligible denominator, and the higher conversion rate on that self-selected segment makes the premium economically justified.
“So what do I actually have to do?”
Tell us the business and give us the account. We build the conversion tracking and prove it works with a test conversion before any ad runs, then launch the campaigns your budget can feed.
After that, changes happen inside a written mandate, so you know in advance what gets changed without asking and what never gets touched. A report on the 1st, a budget email around the 23rd, and nothing else unless it needs your yes or no.
How it works13. Performance Max: Which Levers Exist and Which Are Illusions
Performance Max is not a standard campaign with automation layered on top. It is an automation-first system that accepts inputs and constraints, not instructions. Every lever in the table below must be evaluated as one of three things: a genuine delivery constraint (the platform will respect it), a signal (the platform will weight it but may ignore it), or a reporting view (it tells you what happened but gives you no control). Advertisers who treat signals as constraints overpay for inventory they cannot actually direct; advertisers who ignore reporting views miss the only evidence available about where spend is going.
The table uses six columns. Because the subject requires more detail on certain levers, a second table follows covering the same levers’ reporting availability and what the reader should use as a proxy where native reporting does not exist.
| Lever | Exists in PMax? | What it actually moves | Use it when | Do not use it when | What the platform still decides |
|---|---|---|---|---|---|
| Asset groups | Yes | Creative relevance per theme; asset-level diagnostics | Offers, landing pages or products genuinely differ by group | Splitting for reporting only—creates noise without signal | Which assets show, on which surface, to which user |
| Audience signals | Yes (signal only) | Faster convergence; better early conversion quality | You have a Customer Match list of 1,000+ matched users or strong RLSA data | List is under minimum size; tracking is broken | Whether delivery stays within or expands beyond the signal |
| Search themes | Yes (signal only) | Incremental query discovery on Search and Shopping | Feed or landing-page data is thin and campaign is new | Mature campaign with strong conversion history—themes add noise | Which queries actually trigger; match behavior |
| Listing groups | Yes (constraint) | Which products are eligible for Shopping inventory | You need to exclude low-margin or out-of-stock SKUs | Excluding products to force non-Shopping delivery—it does not work that way | Bid per product; channel allocation for eligible products |
| Campaign-level negatives | Yes (constraint) | Blocks matched queries on Search and Shopping inventory | Irrelevant branded terms, competitor exclusions, known waste queries | Adding broad negatives that block high-converting variants | Query matching beyond explicit negatives |
| Account-level negatives | Yes (constraint) | Cross-campaign query block on Search and Shopping | Terms that must never trigger any campaign (employment, unserved locations) | High-volume terms where exclusion might suppress good traffic in other campaigns | Everything not covered by explicit negatives |
| Lever | Exists in PMax? | What it actually moves | Use it when | Do not use it when | What the platform still decides |
|---|---|---|---|---|---|
| Brand exclusions | Yes (constraint) | Blocks brand-related Search and Shopping queries; covers variants and misspellings | You run a separate brand Search campaign with better economics | No separate brand campaign exists—exclusion just suppresses brand traffic entirely | Non-Search brand appearances (video, Display placements) |
| Final URL expansion | Yes—on by default | Landing-page selection; can generate additional text assets from the destination page | All site pages are conversion-ready and policy-compliant | Some pages lack conversion tracking, are under construction, or are compliance-restricted | Which page is chosen for each auction; asset text generated |
| Budget | Yes (constraint) | Total daily spend ceiling; channel mix shifts with budget level | Marginal CPA is below target and campaign is conversion-volume constrained | CPA is above target—more budget does not fix a quality problem | How budget distributes across channels and time of day |
| Target CPA / ROAS | Yes (constraint) | Efficiency-volume trade-off; loosening target increases impressions and conversions | Conversion volume is above 30/month and tracking is clean | Under 30 conversions/month—target changes cause learning instability[4] | Auction-time bids; channel and query selection within target |
| Feed quality | Yes (direct input) | Shopping eligibility, title relevance, image quality, price competitiveness | Always—feed quality is the strongest lever for Shopping inventory in PMax | No Shopping inventory in the campaign (irrelevant lever) | Which products surface in which auction; bid per product |
| Search + PMax query arbitration | Partial—via brand exclusions and negatives | Which campaign serves branded vs. non-branded queries | Search lost IS (budget) is low and brand CPA differs materially from non-brand | Brand exclusions are already in place—adding more controls creates conflict | Exact query routing when neither campaign is explicitly excluded |
Reporting you do not get—and what to use instead
Search impression share is not reported as a Search-only metric for Performance Max; the platform reports impression share across Search and Shopping combined.[6] Click share is available. There is no keyword-level Quality Score. Asset performance labels (Best/Good/Low) are relative diagnostic ratings, not causal incrementality data. Search-term reporting exists but is subject to privacy thresholds and aggregation—it is not equivalent to the full query log available in a standard Search campaign.[12] Channel-level reporting shows where conversions are attributed, not whether those conversions are incremental. There is no placement-level bid control. Use click share as a proxy for competitive coverage, and use the channel performance report as a directional spend allocation view rather than a bidding tool.
Failure mode: Raising a Performance Max target CPA by 40% in one edit to chase volume resets the learning period—expect 2–4 weeks of volatile CPA and unreliable spend distribution before the system restabilizes; during that window, before-and-after comparisons are meaningless.[4]
When to pull it, and when not to
Pull Performance Max levers when: The account has at least 30 conversions per month with clean tracking; feed quality issues have been resolved (all products approved, titles front-loaded, GTINs submitted); a brand Search campaign with explicit negatives or brand exclusions is in place to protect branded economics; and you are prepared to wait 14–21 days after any target or budget change before reading results.
Do not pull Performance Max levers when: Conversion tracking is miscounted or primary conversion actions include micro-conversions (page views, time on site) that inflate volume—the system will optimize toward noise. Do not add search themes when the campaign already has strong conversion history and a healthy feed; themes are a cold-start lever, and adding them to a mature campaign can dilute query relevance. Do not interpret asset-group performance differences as proof of causal lift without an experiment—delivery mix differences between groups explain most observed performance variance.
Worked example (illustrative)
Brand exclusion economics in a retail PMax campaign
- Setup: A Dallas outdoor-gear retailer running one Performance Max campaign at $14,000 per month in Q2 2026, with no separate brand Search campaign and no brand exclusions. Average CPA: $58. Brand queries (identified via search-term report) account for an estimated 35% of Search/Shopping conversions at a CPA of $22; non-brand conversions run $78 CPA against a target of $65.
- Numbers: Total monthly conversions: 241. Brand-attributed conversions: 84 at $22 = $1,848 spend. Non-brand: 157 at $78 = $12,246 spend. Blended CPA: $14,094 ÷ 241 = $58. Creating a separate exact-match brand Search campaign at $1,848 budget and adding brand exclusions to PMax reallocates that $1,848 to a campaign designed for branded queries. Non-brand PMax now has $12,246 allocated to 157 conversions—CPA unchanged at $78. But the brand campaign’s $1,848 at a $22 CPA generates 84 conversions with explicit keyword control, negative-matching competitor terms, and exact brand ad copy. Total spend unchanged; brand CPA visible and independently controllable.
- Decision: Create a brand Search campaign; apply brand exclusions to Performance Max via the Brand exclusions setting (not keyword negatives, which miss variants). Set brand campaign daily budget to $62 ($1,848 ÷ 30).
- Why: Brand exclusions block brand-related Search and Shopping queries including misspellings and variants from PMax, ensuring the lower-CPA brand traffic is served by a campaign with explicit quality controls and separate reporting, rather than subsidizing the PMax blended average.[12]
14. Demand Gen, Video and Display: Levers on the Demand Side
Demand Gen, Video and Display campaigns do not capture intent that already exists—they create or accelerate it. That distinction matters for every measurement decision: a Search or Shopping campaign can be held to a direct CPA because the user arrived with declared intent; an upper-funnel campaign serves users who have not yet searched, and attributing a last-click or even a data-driven conversion to it overstates its causal role. The honest measurement caveat must precede any evaluation of these levers.
The realistic role of upper-funnel spend in improving Search results
The mechanism by which Demand Gen and Video improve Search performance is indirect: exposure to a video or Discovery ad increases brand search volume, which raises branded Search impression share (numerator rises because more users now search the brand name) and lowers branded CPA (high-intent branded searches convert at 2–4× the rate of non-branded queries). There is no published Google figure quantifying this lift universally; there is no published figure for this lever and the ranges below are reasoned from attribution studies and auction mechanics. A realistic planning assumption is that sustained Demand Gen or YouTube reach at 3+ frequency over 30 days can lift branded search volume by 8–20% for a mid-sized advertiser spending $3,000–$8,000 per month on upper-funnel—but this is a reasoned range, not a documented benchmark, and the effect is undetectable unless branded Search campaigns are separated from non-branded before the test begins.
Campaign type levers and what each one can and cannot control
| Campaign type | Primary creative lever | Audience control mode | CPA accountability | Measurement caveat | Wrong next dollar when |
|---|---|---|---|---|---|
| Demand Gen | Image + video mix; headline/description copy; CTA button | Targeting (expansion optional) | Can be held to Target CPA if 30+ conversions/month exist; expect CPA 1.5–3× Search | View-through conversions inflate attributed CPA; use last-click as a floor check | Non-brand Search IS is below 50% and lost IS (rank) is above 25%—fix intent capture first |
| Video (skippable in-stream) | First 5 seconds; hook; audio-off comprehension; CTA overlay | Targeting | Not reliably holdable to direct CPA; use CPV, view rate, brand lift as primary KPIs | Conversion credit mostly view-through; incrementality requires geo or audience holdout | Account has no remarketing audience; brand awareness is not the constraint on conversion volume |
| Display | Image size variety; headline; description; logo | Targeting or Smart Display (automated) | Can hold to Target CPA in Smart Display with 50+ monthly conversions; Display CPA typically 1.5–2× Search[3] | View-through window inflates attributed conversions; use a 1-day or 0-day view-through window for conservative measurement | Budget is under $5,000/month total—Display spend below ~$800/month rarely generates enough impressions to stabilize Smart Display bidding |
Creative levers inside each campaign type
Demand Gen: Supply all five headline variants (40-character limit) and all five descriptions (90-character limit), plus at minimum one landscape image (1.91:1), one square image (1:1), and one portrait image (4:5). Include at least one video asset when possible—Demand Gen with video assets typically receives more eligible placements on YouTube inventory than image-only campaigns. Audience targeting in Demand Gen operates closer to a hard boundary than Performance Max signals, but enabling optimized targeting allows Google to expand beyond the selected audience when conversion probability is higher elsewhere.[12]
Video: The first 5 seconds must communicate the brand or offer without audio, because 70%+ of YouTube mobile impressions are watched with sound off. Skippable in-stream ads are skipped by most users at the 5-second mark; a 15–20 second non-skippable bumper variant increases forced exposure but cannot carry a CTA. Use vertical (9:16) formats for YouTube Shorts inventory; without a vertical asset the campaign cannot access Shorts placements. Use a custom intent or in-market audience targeting, not broad demographic targeting, when the objective is conversion influence rather than pure reach.
Display: Responsive display ads require: up to 15 images (at minimum one 1.91:1 and one 1:1), up to 5 logos, up to 5 headlines (30 characters), one long headline (90 characters), up to 5 descriptions (90 characters). Display CTR benchmarks run 0.39–0.60% across most industries, with Real Estate an outlier at 1.08%.[3] Display CPC runs $0.44–$0.86 across most industries.[3] Display is a volume-at-low-cost channel, not a high-intent channel; holding it to a Search-equivalent CPA will starve it of spend before it can generate enough data to optimize.
The account conditions under which upper-funnel spend is the wrong next dollar
Upper-funnel spend is the wrong next dollar when: Search lost IS (budget) is above 15%—meaning the account cannot fund the intent-driven auctions it is already eligible for and upper-funnel spend will grow awareness that Search cannot then capture; non-brand Search CPA is above target by more than 30% and the root cause has not been diagnosed (adding awareness spend on top of a broken conversion funnel multiplies wasted spend); the account has fewer than 30 non-brand Search conversions per month, meaning there is no stable baseline against which to measure a branded search lift; or the landing page conversion rate is below 1% on mobile—awareness-driven traffic arriving at a broken page produces no incremental conversions and no measurable lift in any metric.
Failure mode: Setting Display or Demand Gen Target CPA equal to the Search campaign’s CPA before the upper-funnel campaign has accumulated 50 conversions causes constant learning-period churn—the bidding system cannot stabilize, spend oscillates between near-zero and daily-budget-cap in alternating weeks, and the campaign never exits the learning period.
When to pull it, and when not to
Pull Demand Gen, Video or Display levers when: Non-brand Search impression share is above 60% (so intent-capture is reasonably saturated and incremental demand creation makes sense); conversion tracking includes view-through windows set conservatively (1-day or off) so CPA is not inflated; the brand has a creative asset library with at least three distinct message angles and multiple image ratios; and the account has a separate brand Search campaign to capture any lift in branded query volume that the upper-funnel activity generates.
Do not pull these levers when: Search lost IS (budget) is above 15%—every dollar in upper-funnel spend that could have funded a Search impression in an already-eligible auction is a worse economic trade; conversion tracking attributes view-through conversions at a 30-day window, which will make the upper-funnel campaign appear far more efficient than it is and corrupt bidding in any other Smart Bidding campaign sharing the same conversion actions; or the campaign objective is to reduce CPA this quarter—upper-funnel effects on CPA are indirect, delayed by 4–8 weeks, and difficult to isolate without a holdout test.
Worked example (illustrative)
Demand Gen economics check before committing budget
- Setup: A San Francisco HR-software company spending $9,500/month on non-brand Search in Q3 2026 at a $116 CPA[3] with a $140 target. Non-brand Search impression share: 64%. Brand Search generates 18 conversions/month at a $31 CPA from a $558 brand budget. The team is evaluating adding $2,000/month in Demand Gen to lift brand search volume.
- Numbers: Current brand Search: 18 conversions at $31 = $558 spend. If Demand Gen at 3+ frequency over 30 days lifts branded query volume by 12% (low end of the 8–20% reasoned range), that is 18 × 1.12 = 20.2 conversions from brand Search—2.2 incremental brand conversions at $31 each = $68 incremental brand value. The $2,000 Demand Gen budget must also produce its own direct conversions at a CPA below $140 to break even. At a Demand Gen CPA of 1.5× Search ($116 × 1.5 = $174), $2,000 buys 11.5 direct conversions—above the $140 target. At 2× Search CPA ($232), $2,000 buys 8.6 conversions at $232 each—well above target. The break-even requires Demand Gen CPA below $140, which the benchmark range makes unlikely without a proven offer and strong creative.
- Decision: Do not launch Demand Gen at $2,000/month yet. Instead, raise the non-brand Search budget by $2,000 (from $9,500 to $11,500); at the current $116 CPA that adds 17 conversions/month at a known CPA inside the $140 target. Revisit Demand Gen when non-brand Search impression share exceeds 75% and the marginal Search CPA rises above $130.
- Why: When Search lost IS (budget) is non-trivial and Search CPA is inside target, funding intent-capture at a known CPA dominates funding demand creation at an uncertain and likely higher CPA; the opportunity cost of the upper-funnel dollar is a Search conversion the account could have bought directly.
15. Shopping and Merchant Center: Feed Levers That Move Impression Share
For a retail account running Shopping or a Performance Max campaign with a product feed, the feed itself is the single biggest lever available. Unlike Search campaigns where ad copy, keywords and bids all share influence roughly equally, Shopping eligibility is determined almost entirely by whether the feed gives Google enough correct, complete data to match a product to a query. A product with a poor title, missing GTIN and low-quality image can be outbid from almost any auction regardless of how high the campaign’s CPC bid is set, because Google uses feed data to determine relevance before it evaluates price. The practical implication is that a feed audit should precede any bidding change in a Shopping or Performance Max retail account that is underperforming on impression share.
Shopping impression share is calculated from Search Network traffic only and excludes Display and Search Partners.[6] It is not the same metric as Search impression share in a text-ad campaign, and it is not reported for Demand Gen or Video. For Performance Max, Google calculates impression share using Search plus Shopping impressions combined, so a pure Shopping impression share view requires either a standard Shopping campaign or a Performance Max report segmented by channel.[6]
Failure mode: An advertiser who raises bids or increases budget on a Shopping campaign with widespread feed errors will buy more impressions on a broken eligibility set, inflating cost with no proportional lift in conversions, while the root disapprovals and relevance deficits remain untouched.
| Feed or Merchant Center action | What it moves | Realistic size of effect | Use it when | Do not use it when | Where you do it |
|---|---|---|---|---|---|
| Rewrite product titles (brand + type + key attributes front-loaded) | Shopping impression share, CTR, conversion rate | 10–30% impression-share gain for affected products; CTR gain varies by category | Titles are generic, truncated or missing brand/model; Ad relevance is weak | Titles already front-load brand, model and key attributes accurately | Primary feed or supplemental feed; check Merchant Center → Products → All products |
Add or correct product_type taxonomy (3+ levels deep) |
Eligibility for structured product groups; campaign subdivision accuracy | Unlocks bidding by subcategory; indirect impression-share gain on mis-filed products | Product groups in campaigns show “All products” with no subdivision | Taxonomy is already 3+ levels and campaign subdivision is in use | Primary feed or feed rules; Merchant Center → Settings → Data sources |
Submit accurate google_product_category (2–3 levels minimum) |
Google’s ability to classify products; eligibility on category-restricted surfaces | Removes category mismatch disapprovals; improves relevance scoring | Category shows as top-level only or is missing for branded products | Category is already specific and matches the product accurately | Primary feed; Merchant Center → Products → Needs attention |
| Add correct GTINs and brand/MPN for all products that have them | Eligibility on GTIN-required surfaces; product matching quality | Resolves GTIN-related disapprovals; can recover impression share lost to eligibility gaps | Products → Needs attention shows GTIN warnings or missing-identifier issues | Products are genuinely custom or handmade with no manufacturer GTIN | Primary feed; Merchant Center → Products → Needs attention → filter by GTIN issue |
| Replace low-resolution or promotional-overlay images | CTR, image-disapproval rate, ad eligibility | Removes image disapprovals; CTR lift is directional, not a fixed percentage | Products → Needs attention shows image-quality warnings; CTR is below category benchmark | Images are already clean, high-resolution product shots with no overlays | Primary feed [image_link]; Merchant Center → Products → Issues → Image |
| Fix price or availability mismatches between feed and landing page | Disapproval rate, eligible impression share, landing-page experience | Resolves price/availability disapprovals immediately on reprocessing; can recover large eligibility blocks | Products → Needs attention shows price or availability mismatch errors | Feed and page prices are already synchronized via automated API or structured data | Merchant Center → Products → Needs attention; automate via Content API or structured data markup |
| Submit promotions through Merchant Center promotions tool | CTR via promotion annotation (“Special offer”) in Shopping ad | Promotion annotation differentiates ad from competitors; CTR effect varies by discount depth and category | A genuine, time-limited discount exists and the landing page shows the same price | No real discount exists or landing-page price does not match | Merchant Center → Marketing → Promotions; verify via Google Ads → Segment → Click type |
Add sale_price and sale_price_effective_date |
Price-competitiveness rank; sale annotation eligibility | Improves price-comparison ranking where sale price is below market; enables strikethrough annotation | A temporary sale is running and the regular price is set in [price] |
Sale period has ended; landing-page and checkout price do not reflect the sale price | Primary feed attributes; Merchant Center → Analytics → Pricing → Price competitiveness |
| Assign custom labels (margin tier, best-seller, seasonal, price band) | Campaign bidding precision; budget allocation by commercial priority | Enables different bid strategies by label group; indirect CPA and ROAS improvement | All products are in one bid group and margin or value varies materially across catalog | Catalog is small and uniform in margin; labeling would create groups too thin to bid reliably | Primary or supplemental feed [custom_label_0–4]; subdivide in campaign product groups |
| Use a supplemental feed to patch titles, labels or sale prices at scale | All the metrics of the attributes it corrects, without rebuilding the primary feed | Same as the corrected attribute; supplemental feed changes process within the normal feed refresh cycle | Primary feed is managed by a third party and cannot be edited directly in time | Primary feed can be edited directly; supplemental feed values conflict with primary feed values | Merchant Center → Settings → Data sources → Supplemental sources → Add supplemental source |
| Resolve all disapproved items at Merchant Center → Products → Needs attention | Eligible impression share; total active product count | Every disapproved item is 0% impression share; resolving 20% of catalog disapprovals can lift share proportionally | Disapproval rate is above 5% of active products by impression count | Disapprovals are on low-priority, low-margin products that are not worth advertising | Merchant Center → Products → Needs attention; filter by issue type; fix in feed |
Price competitiveness as the ceiling constraint
Even a perfect feed cannot overcome a price that is materially above the market for a commodity product. The Merchant Center price-competitiveness report (Merchant Center → Analytics → Pricing → Price competitiveness) shows how each product’s price compares to other merchants. Where a product is 15% or more above the median competitive price, the ceiling on click share and conversion rate is set by the offer, not the feed or the bid. The correct response is to either reduce the price, improve shipping and returns terms, add a promotion, or accept that the product will not be competitive in direct-comparison Shopping auctions and redirect budget to products where the offer is competitive. No feed optimization or bidding change will move a materially uncompetitive product past this ceiling.
Worked example (illustrative)
Feed title rewrite recovering lost Shopping impression share
- Setup: A Dallas home-goods retailer spending $6,200 a month on a standard Shopping campaign, Shopping impression share of 38%, Search lost IS (rank) of 44%, and Search lost IS (budget) of 8%. The Merchant Center Products → Needs attention tab shows no active disapprovals, but 60% of product titles begin with an internal SKU code rather than brand and product type.
- Numbers: The campaign runs 1,400 active products at an average CPC of $0.68[1] and a shopping conversion rate of 2.1%[1], yielding roughly 9,118 clicks and 191 conversions per month at a CPA of $32.46. After rewriting titles on the top 400 products by impression count to front-load brand, product type and key attributes, Merchant Center reprocesses the feed within 48 hours. Over the following 28 days, Shopping impression share rises from 38% to 51%, a 13-point gain driven by improved relevance matching (the numerator rises; the denominator is unchanged because targeting did not change). Clicks rise from 9,118 to approximately 12,000, a 32% gain, and at the same 2.1% conversion rate that is 252 conversions. Spend rises from $6,200 to roughly $8,160 at the same average CPC, adding $1,960 of spend for 61 additional conversions at a marginal CPA of $32.13, below the account target of $38.
- Decision: Rewrite the
[title]attribute for the top 400 products by prior-period impressions to follow the structure: Brand + Product Type + Key Attribute + Size/Color. Deploy via supplemental feed to avoid disrupting the primary feed schedule. - Why: Google uses the title as the primary relevance signal for Shopping query matching; a title that begins with an SKU produces weak relevance scores and reduces the number of eligible auctions even when the bid is competitive, so fixing the title expands the eligible pool (the denominator of impression share stays the same, but the campaign wins more of it because relevance improves).
When to pull it, and when not to
Pull feed levers when: Shopping impression share is below 50% and Search lost IS (rank) is above 25% for affected products; Merchant Center → Products → Needs attention shows active disapprovals, GTIN warnings or image issues on products that account for more than 10% of catalog impressions; titles begin with SKUs, internal codes or generic descriptions; the price-competitiveness report shows affected products within 10% of the market median; or the campaign has sufficient budget (Search lost IS (budget) below 10%) but impression share is still low, indicating the constraint is relevance, not spend.
Do not pull feed levers when: The feed is already optimized (titles are specific, identifiers are correct, images are clean) and the remaining impression-share gap is caused by budget exhaustion (Search lost IS (budget) above 20%), in which case the binding constraint is funding, not feed quality and no feed change will recover those impressions. Also do not invest in feed work for products where the price-competitiveness report shows the product is more than 20% above the market median price, because relevance improvements will drive clicks that will not convert at an acceptable CPA; fix the offer or exclude those products first.
16. Landing Page and Conversion Levers: The Cheapest Wins in the Account
A landing-page or conversion-rate improvement reduces cost per conversion without touching the auction. That makes it structurally cheaper than buying the same CPA improvement through impression share, because impression share improvements require more spend to hold the same economics while a conversion-rate improvement improves economics at the same spend. The arithmetic is straightforward: at $5.26 average CPC[1] and a 7.52% conversion rate[1], the cost per conversion is $5.26 ÷ 0.0752 = $69.95. If the conversion rate rises to 9.0%, the cost per conversion falls to $5.26 ÷ 0.09 = $58.44, an $11.51 saving per conversion requiring zero additional spend and no bid change. Achieving the same $11.51 saving by reducing CPC would require cutting average CPC from $5.26 to $4.47, a 15% reduction that is rarely achievable without also reducing impression share.
The ceiling on what landing-page optimization can deliver is the offer itself. A landing page that correctly represents a poor offer, a confusing pricing structure or an uncompetitive product cannot be optimized past the ceiling set by the offer. Identify whether the problem is the page or the offer before spending engineering time on CRO.
Failure mode: An account running Smart Bidding toward a conversion action that is misconfigured—counting page views as conversions, double-counting form submissions or firing on a URL that is not a genuine completion—cannot be corrected by any bidding lever, budget increase or quality improvement. The bidding system optimizes toward the signal it receives; a bad signal produces a campaign that optimizes efficiently toward the wrong outcome, spending confidently in the wrong direction.
Page speed
Google’s research reported that as mobile page load time rises from 1 second to 5 seconds, the probability of a mobile visitor bouncing increases by 90%, and that a 1-second delay in mobile page load can reduce mobile conversions by up to 20%.[5] These are directional benchmarks; the actual conversion loss depends on baseline speed, device split, page type and audience intent. Measure current Core Web Vitals at PageSpeed Insights or Search Console before and after any change. A page loading in under 2 seconds on mobile is a reasonable working threshold; above 4 seconds, speed is likely a material conversion drag on mobile traffic. The conversion-rate effect appears within days of deployment; the Quality Score landing-page-experience component may take longer to reflect the change.
Message match
Message match means the landing page headline and above-the-fold content directly confirm the ad’s specific promise. A user who searches “emergency HVAC repair Dallas” and clicks an ad reading “24/7 Emergency HVAC Dallas” expects to arrive at a page that confirms emergency availability, service area and the next step, not a homepage about the company’s history. Poor message match increases bounce rate, reduces conversion rate and can lower Quality Score landing-page experience to Below average. The fix is usually to create a dedicated landing page per ad group theme or at minimum to use dynamic keyword insertion or parameter-based headline customization. The conversion-rate improvement from moving from a generic homepage to a matched landing page is typically 20–60% relative where the original experience was severely mismatched; there is no published Google figure for this, and the range reflects the severity of the original mismatch.
Form length and lead quality
Reducing form fields increases submission rate but can reduce lead quality. The correct test metric is not form submissions but qualified leads or downstream revenue per click. Removing three optional fields from a seven-field form commonly lifts raw submission rate by 15–30% (there is no published Google figure; this is a CRO industry range from multiple published A/B test compilations and should be verified with the account’s own experiment). Always measure the qualified lead or sale rate on the shorter form before declaring a win, because a form that generates twice as many submissions at half the close rate has not improved the account’s economics.
Conversion tracking quality
Before any optimization lever is pulled, verify that the account is measuring the right thing. In Google Ads, navigate to Goals → Conversions → Summary and check every primary conversion action. Confirm that Count is set correctly (Every for purchases, One for leads), that the conversion window matches the realistic sales cycle, that the conversion action is firing on a genuine completion URL and not on a thank-you page that is also reached by organic traffic, and that enhanced conversions or the Google tag is configured. If the conversion volume reported in Google Ads does not reconcile within roughly 10% of the actual business outcome (actual sales, booked appointments, qualified leads), the bidding system is optimizing against a flawed signal and no bid strategy change will correct the underlying problem.
Worked example (illustrative)
Conversion-rate improvement versus impression-share improvement: cost comparison
- Setup: A Portland B2B software account spending $12,000 a month on Search at a $5.47 CPC[1], 4.9% conversion rate[1] and Search impression share of 55%. The account target is $115 CPA against an actual CPA of $5.47 ÷ 0.049 = $111.63. The marketing team is debating whether to raise the budget to push impression share to 70% or to invest in landing-page redesign.
- Numbers: Current state: 12,000 ÷ 5.47 = 2,193 clicks, 2,193 × 0.049 = 107 conversions, CPA $111.63. Option A — push impression share from 55% to 70% by raising budget: a 15-point share gain on a $5.47 CPC account requires roughly 27% more spend, adding $3,240 to reach approximately $15,240/month. At the same 4.9% conversion rate that produces 139 conversions at a CPA of $109.64, saving $2 per conversion at a cost of $3,240 more per month. Option B — improve conversion rate from 4.9% to 6.5% via landing-page redesign (message match, form reduction, page speed from 4.2 to 1.8 seconds): 2,193 clicks × 0.065 = 143 conversions at the same $12,000 spend, CPA = $83.92, saving $27.71 per conversion at zero additional media spend. Option B delivers 36 more conversions per month at $27.71 lower CPA than Option A, at no additional spend and no learning-period reset.
- Decision: Invest in landing-page redesign first: fix page load time to under 2 seconds, replace the homepage destination with a campaign-specific landing page matching the ad’s exact offer, and reduce the contact form from 9 fields to 5. Reassess impression share after 60 days of stable new conversion data.
- Why: A conversion-rate improvement reduces cost per conversion by increasing the denominator (conversions) without changing the numerator (spend), while an impression-share improvement increases both spend and conversions proportionally, leaving cost per conversion nearly unchanged and requiring permanent additional budget to maintain.
When to pull it, and when not to
Pull landing-page and conversion levers when: The account’s conversion rate is more than 20% below the industry benchmark for the category[1]; Google Analytics or the landing-page analytics tool shows a bounce rate above 65% from paid search traffic; PageSpeed Insights reports mobile load time above 3 seconds; the ad’s offer and the landing-page headline describe different things; the Goals → Conversions → Summary view shows primary conversion actions with implausible volume (too high or too low relative to actual business outcomes); or the CPA is above target despite Search impression share and Click-through rate both being healthy.
Do not pull landing-page levers when: The conversion tracking setup has not yet been verified — fixing a misconfigured conversion action must come first, because a page improvement on a campaign with a bad signal will not produce reliable data and may reset Smart Bidding learning. Also do not invest in CRO when the offer itself is uncompetitive (price too high, shipping too slow, guarantee weaker than competitors’), because no page design improvement can overcome a structurally inferior offer; fix the offer first. Finally, do not declare a landing-page test a winner until it has accumulated at least 100 conversions per arm; a test called on fewer than 50 conversions per variant has insufficient power to detect realistic effect sizes.
17. Sequencing and Realistic Expectations: What Is Reasonably Possible in 30, 60 and 90 Days
The most common optimization mistake is changing multiple levers simultaneously, then being unable to attribute any improvement or degradation to a specific action. Each major change — bid strategy switch, target CPA revision above 20%, keyword match-type restructure, landing-page replacement, feed overhaul — should be treated as a separate test with a predefined evaluation period. The minimum conversion volume before a change can be judged is 30 conversions in the evaluation period for a directional read, and 100 conversions for a decision to make a further change with reasonable confidence.[4] Below 30 conversions in the window, variance dominates and any observed movement is as likely to be noise as signal.
The table below assumes an account that is already spending, has functioning conversion tracking, and has at least 30 conversions per month. An account with fewer than 30 conversions per month must resolve volume first — usually by fixing conversion tracking, consolidating campaigns or relaxing match types — before the levers below apply.
| Phase | Lever | What to expect | Time to read | How to prove it worked |
|---|---|---|---|---|
| Days 1–30: Diagnose and fix the floor | Verify conversion tracking; resolve disapprovals; fix budget vs. rank diagnosis; add critical negatives | Conversion data becomes reliable; eligible impression share recovers on affected products; irrelevant spend stops | 14–21 days after fixes deploy | Conversion count reconciles with actual business outcomes within 10%; Search lost IS (budget) and lost IS (rank) move in the predicted direction |
| Days 1–30: Diagnose and fix the floor | Landing-page speed and message match for the highest-spend ad groups | Conversion rate improvement of 15–50% relative where mismatch was severe; 0–10% where pages were already specific | 14 days for conversion-rate signal; 21–28 days for Quality Score landing-page experience to update | Conversion rate in Google Analytics rises for paid search sessions on affected pages; CPA falls without a budget or bid change |
| Days 31–60: Improve relevance and bidding | Ad relevance: restructure ad groups, rewrite RSA assets for Below average keywords | Ad relevance status moves from Below average toward Average on affected keywords; CTR may rise 5–15% relative | 14–21 days after new ads accumulate impressions | Ad relevance status improves in Audiences, keywords, and content → Search keywords → Columns → Quality Score; impression-weighted CTR rises |
| Days 31–60: Improve relevance and bidding | Bid strategy target adjustment: raise Target CPA by 10–20% if Search lost IS (rank) is above 20% | Impression share gain of 5–15 points; CPA rises toward the new target, ideally staying below the account ceiling | 14 days after adjustment; allow one full conversion-lag cycle | Search lost IS (rank) falls; Search impression share rises; CPA measured over 14-day post-change window vs. 14-day pre-change window |
| Days 61–90: Scale what is working | Budget increase of 20–30% on campaigns meeting CPA or ROAS targets with Search lost IS (budget) above 10% | Conversion volume rises roughly in proportion to spend; marginal CPA may rise 5–15% as less certain auctions are entered | 14–21 days for delivery to stabilize at new budget level | Incremental CPA (new spend minus baseline spend, divided by incremental conversions) stays below target; Search lost IS (budget) falls |
| Days 61–90: Scale what is working | Feed title and identifier improvements for Shopping; supplemental feed deployment | Shopping impression share gain of 10–30 points for affected products; CTR and conversion-rate effects directional | 21–28 days after Merchant Center reprocesses the feed | Shopping impression share rises in Google Ads campaign view; product-level impressions and clicks rise in Merchant Center Analytics → Products |
Realistic total improvement: competently run versus poorly run account
An account that is already well-structured — conversion tracking is accurate, negatives are in place, bids are on a Smart Bidding strategy with 50 or more conversions per month, landing pages match the ad — has a realistic 90-day ceiling of 10–20% improvement in cost per conversion, achievable primarily through creative testing, incremental bid-target tuning and landing-page CRO. An account that arrives at this process with misconfigured conversion tracking, no negatives, all keywords on broad match with no Smart Bidding, and landing pages pointing to the homepage has a realistic 90-day ceiling of 40–70% improvement in cost per conversion, because the floor was set by avoidable errors rather than competitive constraints. In both cases, the ceiling beyond 90 days is set by the offer, the competitive market and the available search demand — no optimization sequence can manufacture demand that does not exist or win auctions that the account’s economics cannot support.
Never change bid strategy, target CPA or target ROAS, campaign structure, and conversion actions in the same week. Each combination makes it impossible to isolate which change drove the result, prolongs any learning period and can produce a 2–4 week period of elevated CPA volatility that looks like failure but is actually noise from simultaneous resets.[4]
Worked example (illustrative)
90-day sequence for an underperforming B2B lead-gen Search account
- Setup: A Chicago IT services account spending $8,500 a month on Search at a $5.47 CPC[1], Search impression share of 42%, Search lost IS (rank) 38%, Search lost IS (budget) 6%, conversion rate 3.1% (below the B2B Services benchmark of 4.9%[1]), and CPA of $5.47 ÷ 0.031 = $176.45 against a target of $130. Conversion tracking fires on a thank-you page but audits reveal a 22% double-count rate from organic sessions hitting the same URL.
- Numbers: Days 1–30: Fix conversion tracking to fire only on paid-search-attributed sessions; actual conversion rate recalculates to 3.8% and CPA to $5.47 ÷ 0.038 = $143.95, still above target but now a true baseline. Add 47 negative keywords from the Search terms report covering informational and employment queries representing $820 of monthly spend. Days 31–60: Rewrite RSA assets for 6 ad groups where Ad relevance reads Below average; 21 days later CTR rises from 4.2% to 5.1%, a 21% relative gain. Raise Target CPA from $143 to $160 (12% increase); 14 days later Search lost IS (rank) falls from 38% to 24% and Search impression share rises from 42% to 55%. CPA in the measurement window is $148. Days 61–90: Redirect top-spend ad groups from the homepage to a campaign-specific landing page; conversion rate rises from 3.8% to 5.3% over 21 days. CPA falls to $5.47 ÷ 0.053 = $103.21. Increase budget by 25% to $10,625; incremental spend of $2,125 at a marginal CPA of $112, below the $130 target.
- Decision: Execute in three distinct phases with a 14-day hold after each change before the next lever is pulled. Total 90-day improvement: CPA from $176.45 (inflated by tracking error) to $103.21 on a sound tracking basis, a 42% reduction. No single lever achieved this; each lever operated on a different constraint in sequence.
- Why: Fixing conversion tracking first prevented the bidding system from optimizing against false signal; eliminating wasted spend on negative keywords freed budget for better auctions; improving ad relevance lifted CTR before the bid target was raised; landing-page improvement raised conversion rate before budget was scaled, so the incremental spend entered at a better economics level than the baseline.
When to stop optimizing
Stop making changes and hold the account stable when: Search impression share is above 70%, Search lost IS (rank) is below 10%, Search lost IS (budget) is below 5%, the CPA or ROAS is within 5% of the target over a rolling 30-day window, and no single lever identified in sections 7 through 16 shows a testable gap. At this point the account is at or near its optimization ceiling given current targeting, offer and budget, and further changes are more likely to introduce noise than improvement. The correct action is to hold, monitor and collect data, then revisit when competitive conditions, offer, budget or seasonal patterns change materially.
Stop advertising on this channel for this offer entirely when: the account has run for at least 90 days with competent conversion tracking, appropriate negatives and a tested landing page, the CPA remains more than 50% above the target despite pulling the levers in section 17’s sequence, and the price-competitiveness report or customer feedback indicates the offer is not competitive in the market. In that condition no optimization lever available within Google Ads can close the gap, because the constraint is the offer or the unit economics, not the campaign. Continuing to spend is buying data that confirms a known result.
18. What Changed Recently
Research available for this article returned two dated, authoritative change notices from Google Ads Editor release notes. Neither falls strictly within a 30-day window relative to publication, and no other officially dated platform changes were available to report with confidence.[13]
Google Ads Editor 2.13 (October 27, 2026): Added editor support for Performance Max customer-retention goals, Demand Gen Maps placements, and Shopping AI Max controls. Also removed the ability to create new Video Action campaigns via the editor.[14] These are editor-side capability additions, not account-level auction or bidding behavior changes.
Google Ads Editor 2.14 (November 13, 2026): Reported default activation of AI Max, text-asset automation, and Final URL expansion for newly created Search campaigns; permanent enablement of Local Inventory Ads for Shopping campaigns; and new Demand Gen asset-optimization support.[15] The Final URL expansion default is the most operationally significant item for Search advertisers: accounts creating new Search campaigns after this editor version may have Final URL expansion active by default, meaning Google can serve landing pages other than the specified final URL. Advertisers who require specific landing-page control should verify the setting at campaign level in the Google Ads interface under Campaign settings → Additional settings → Final URL expansion and add URL exclusions where necessary.
No other platform changes — to bidding logic, auction mechanics, Quality Score calculation, impression share reporting, Merchant Center feed processing, or Performance Max channel controls — were available from dated, authoritative sources for the period covered. Where changes in this area affect the levers described in sections 7 through 16, the most reliable source is the Google Ads announcements page at support.google.com/google-ads/announcements/9048695 and the Merchant Center release notes.
Related reading
- Building a Google Ads Strategy
- Responsive Search Ads: Headlines, Pinning and Ad Strength
- Google Search Ads: How Search Campaigns Work
- Google Ads Seasonality Adjustments and Promotions
Or skip the work and get professional Search campaign setup, without an agency retainer.
References
- [1] https://support.google.com/google-ads/answer/12732144?hl=en-GB support.google.com
- [2] https://searchengineland.com/google-ads-costs-keep-rising-but-conversion-rates-improved-in… searchengineland.com
- [3] https://adbacklog.com/blog/google-ads-benchmarks-per-industry-2025 adbacklog.com
- [4] https://support.google.com/google-ads/answer/7634668?hl=en support.google.com
- [5] https://support.google.com/google-ads/answer/1722084?hl=en support.google.com
- [6] https://support.google.com/google-ads/answer/6167118 support.google.com
- [7] https://support.google.com/google-ads/answer/6297?hl=en-EN support.google.com
- [8] https://www.support.google.com/google-ads/answer/1722122 www.support.google.com
- [9] https://support.google.com/google-ads/answer/6167118?hl=en-GB&ref_topic=24937&visit_id=638… support.google.com
- [10] https://support.google.com/google-ads/answer/13738235?hl=en support.google.com
- [11] https://support.google.com/google-ads/answer/17035334?hl=en support.google.com
- [12] https://support.google.com/google-ads/answer/10215?hl=en support.google.com
- [13] https://support.google.com/google-ads/answer/6167118?visit_id=638194707463603871-131005457… support.google.com
- [14] https://support.google.com/google-ads/editor/answer/17136249?hl=en support.google.com
- [15] https://support.google.com/google-ads/editor/answer/18835571?hl=en support.google.com

