Performance Max Campaigns

Last updated: October 4, 2026

1. Executive Summary: Five Principles for Performance Max in 2026

Performance Max (PMax) has matured into Google’s primary campaign type for full-funnel, cross-channel conversion optimization. Run well, it consolidates Search, Shopping, YouTube, Display, Discover, Gmail, and Maps inventory under a single Smart Bidding strategy and delivers measurable efficiency gains. Run carelessly, it wastes budget on brand queries, low-intent placements, and AI-assembled creative that no human approved. The five principles below govern every recommendation in this guide.

• Principle 1 – Feed the machine first-party data. Google’s AI is only as good as the signals you provide. Customer Match lists, website visitor audiences, and clean conversion tagging are prerequisites, not optional enhancements.^[1]
• Principle 2 – Let the learning period breathe. PMax requires 2–4 weeks and at least 30–100 monthly conversions before targets stabilize. Changing bids, budgets, or creative during this window restarts learning and compounds instability.^[2][3]
• Principle 3 – Structure for signal clarity, not campaign proliferation. Fewer, broader, clearly themed asset groups outperform dozens of micro-segmented ones because the algorithm needs volume to learn. One campaign per objective; one asset group per meaningful theme.^[4][5]
• Principle 4 – Own your creative, supplement with AI. AI-generated assets reduce serving friction but should not replace first-party video, image, and copy. Achieve Excellent Ad Strength through human-created asset variety before relying on automated generation.^[4][6]
• Principle 5 – Measure with discipline and protect your brand. Use the Insights page, channel-level reporting, and placement segmentation to understand where budget flows. Apply brand exclusions and audience exclusions proactively to prevent cannibalization and brand-safety violations.^[7][8]

2. Benchmarks and Numbers at a Glance

Metric	Typical range or threshold	Applies when	Source
PMax average CPA (cross-industry)	$43.91	Broad cross-industry US data; vendor study, sample size not stated	[4]
Search average CPA (cross-industry)	$53.52	Same dataset as PMax CPA figure above; vendor study	[4]
PMax ROAS – e-commerce (broad range)	3.5×–5.0×	E-commerce accounts; vendor study, sample size not stated	[4]
PMax ROAS – e-commerce (Q3 2026 tracked accounts)	5.91×	25 tracked e-commerce accounts, Q3 2026	[9][10]
PMax ROAS – B2B SaaS	4.36×	B2B SaaS accounts; vendor study, sample size not stated	[4]
PMax average conversion rate	3.42%	Cross-industry PMax data; vendor study, sample size not stated	[11]
PMax average cost per conversion	$72.06	Cross-industry PMax data; vendor study, sample size not stated	[11]
Minimum monthly conversions to run PMax effectively	15–30 conversions/month	Campaign management baseline threshold	[3]
Minimum monthly conversions to add a tCPA target	30–100 conversions/month	Before applying Target CPA to a PMax campaign	[3]
PMax learning period	2–4 weeks / 2–3 conversion cycles	After any significant change to bid strategy, budget, or creative; vendor claim	[2][3]
Minimum recommended daily budget	$50–$100+/day	Entry-level PMax; vendor study, sample size not stated	[12]
CPA reduction from strong audience signals (first 30 days)	15% lower CPA vs. weak-signal launch	When high-quality Customer Match and behavioral lists are loaded at launch; vendor study	[13]
CPA reduction vs. manual bidding (after 6-week learning)	24% lower CPA	After full 6-week Smart Bidding learning period; vendor study	[13]
ROAS lift from custom audience segments as signals	15%–30% ROAS lift	Custom segments built from search terms used as audience signals; vendor study	[14]
New-customer ROAS lift (new customer acquisition goal)	9% ROAS increase; 5% higher new-customer ratio; 7% lower new-customer acquisition cost	Advertisers using PMax new-customer acquisition mode; vendor claim	[5]

3. When to Use Performance Max

Performance Max is the right choice when your account has a single clear conversion objective, enough conversion volume for Smart Bidding to learn from, and creative assets that can span multiple formats and channels. It is not the right choice when you need rigid keyword control, granular negative-keyword lists at launch, or when monthly conversion volume falls below 15.^[3]

Strong use cases

• E-commerce with a Merchant Center feed: PMax subsumes Shopping inventory and pairs feed-based product ads with upper-funnel video and display. The algorithm can optimize simultaneously for revenue and new customers.^[5][12]
• Lead generation at scale: Accounts generating 30 or more qualified leads per month can use tCPA to drive cost efficiency. Accounts below that threshold should run Maximize Conversions without a target first.^[3][12]
• Local and service businesses: PMax can serve across Search, Maps, and Display simultaneously. As of 2026, Local Services Ads are migrating into PMax pay-per-lead campaigns in phases, making PMax the consolidation point for local intent traffic.^[15][16]
• Full-funnel awareness plus conversion: Accounts that want YouTube and Display reach without a separate campaign structure benefit from PMax’s cross-channel allocation.^[9]

When to pause before launching PMax

• Fewer than 15 conversions per month in the account — bidding signals are insufficient.^[3]
• Conversion tracking is incomplete or firing inconsistently — bad input data produces bad optimization.^[3]
• The account requires tight keyword-level negative control that cannot be achieved at the campaign level.
• Brand spend must be separated and capped — ensure brand exclusions are configured before launch.^[8]

4. Campaign Structure and Asset Groups

The governing rule for PMax structure is signal volume over granularity. Splitting budget across many narrow campaigns or dozens of micro-asset groups denies the algorithm the conversion data it needs to learn. Google’s own creative playbook recommends a small number of broad, clearly themed asset groups rather than many narrow ones.^[5][6]

Campaign-level architecture

• One campaign per conversion objective. Mixing lead-gen and e-commerce purchase goals inside a single campaign confuses Smart Bidding. Separate objectives into separate campaigns.^[8]
• Separate campaigns for materially different margin or ROAS targets. A retailer with a 60% gross-margin product line and a 20% gross-margin product line should run separate campaigns so tROAS targets reflect each line’s economics.^[12]
• Avoid overlapping Standard Shopping campaigns. When a linked Merchant Center feed is present, a PMax campaign takes priority over a Standard Shopping campaign for the same products. Run them in parallel only if you intend to test via a formal Experiment.^[9]

Asset group structure

• One asset group per meaningful theme, product category, or audience. Each group should have a clear final URL, a coherent creative set, and audience signals that match the theme.^[4][5][17]
• For retail, organize by product set — not by channel or device — so the algorithm can match the right products to the right queries and audiences.^[12]
• Do not fragment into micro-groups. A group with only a few hundred impressions per week cannot generate enough conversion signals for meaningful learning.^[4][6]
• Aim for Excellent Ad Strength by filling all available asset slots with diverse, non-repetitive creative before considering whether to replace any asset.^[14]

5. Creative and Asset Strategy

Asset quality and variety are the primary levers a marketer controls inside PMax. Google’s AI assembles combinations across surfaces; your job is to maximize the number of high-quality, non-repetitive combinations available. Every asset group should have complete coverage across text, images, and video before launch.^[4][9][10]

Text assets

• Write headlines and descriptions that make sense in any combination — avoid assets that depend on appearing alongside a specific partner asset.^[5]
• Include benefit-focused, non-generic copy. “Free 2-day shipping on orders over $50” is actionable. “Great products at great prices” is not.^[5][6]
• Vary message angle: lead with price, then lead with outcome, then lead with urgency. These variations give the algorithm distinct options for different audience contexts.^[4]

Image assets

• Supply images in all required aspect ratios: 1.91:1 (landscape), 1:1 (square), and 4:5 or 9:16 (portrait/vertical) to maximize serving surfaces.^[4][6]
• Use high-quality product photography against clean backgrounds for Shopping surfaces, and lifestyle or context imagery for Display and Discover.^[6]
• Do not rely on a single image variant. More distinct images create more testable combinations.^[4]

Video assets

• Video is required for serving across YouTube and other video inventory. Omitting video means Google may auto-generate assets from your images and text — a lower-quality fallback.^[2][9]
• Provide videos in horizontal (16:9), square (1:1), and vertical (9:16) orientations to cover all YouTube and Shorts placements.^[4]
• Include at least one video under 30 seconds for non-skippable inventory and at least one video 30 seconds or longer for skippable in-stream inventory.^[10]

Ad Strength and asset quality

• Aim for Excellent Ad Strength. Fill all available slots before replacing any asset. Google advises waiting 2–3 weeks for asset-level performance labels to stabilize before removing assets.^[4][14]
• Use asset performance labels (Best, Good, Low) to guide replacement decisions, not Ad Strength alone — an asset marked Low may still serve well in certain contexts.^[4]
• AI-generated assets should be reviewed and approved before serving where account settings permit. Treat them as a supplement to, not a replacement for, first-party creative.^[2][3][9]

6. Audience Signals and Search Themes

Audience signals and search themes in PMax are hints to the algorithm, not hard targeting constraints. Google AI can and will expand beyond your inputs when it detects stronger conversion likelihood elsewhere.^[4][9] Your objective is to provide the highest-quality signals possible so the algorithm starts in the right direction and shortens its learning period — research suggests strong signals at launch reduce CPA by 15% in the first 30 days and accelerate learning by 25%.^[13]

Priority order for audience signals

• First-party Customer Match lists — purchasers, high-LTV customers, recent cart abandoners, qualified leads. These are the highest-quality signal inputs.^[4][17]
• Website and app behavioral lists — all visitors, product-page visitors, checkout-abandonment audiences from Google tag or GA4 integration.^[1][17]
• Custom segments — built from high-converting search terms, competitor URLs, or apps your customers use.^[4][17]
• Google audience segments — In-market, Affinity, Life Events, Detailed Demographics — use only when they materially match your buyer profile.^[4][17]

Signal quality requirements

• Customer Match lists should contain more than 1,000 active, eligible users to be most useful.^[17]
• Lists must be refreshed within 540 days to remain eligible; stale lists degrade signal quality.^[17]
• Improving Google tag coverage across all conversion pages directly improves both audience list accuracy and conversion reporting.^[3]
• Expect a learning period of up to 2 weeks after adding or changing audience signals before the model fully incorporates them.^[17]

Search themes

• Search themes communicate commercial intent at the asset group level. Choose terms that reflect your highest-value product categories and the problem/solution language your buyers use.^[3][14]
• Themes should be specific enough to be directionally useful but broad enough to let the algorithm identify adjacent queries.^[3]
• Search themes are not exact-match keyword triggers. Do not use them as a substitute for a tightly managed Search campaign when keyword-level control is required.^[3][9]

7. Bidding Strategy: tCPA and tROAS

Smart Bidding in PMax operates on either a conversion-count objective (Maximize Conversions / Target CPA) or a conversion-value objective (Maximize Conversion Value / Target ROAS). Choosing the right strategy depends on conversion volume, value reliability, and where the account is in its learning cycle. Premature targets and over-aggressive values are the most common bidding mistakes in PMax accounts.^[3][12]

Progression framework

Stage	Monthly conversions	Recommended strategy	When to advance
Launch / Learning	< 30/month	Maximize Conversions (no tCPA)	After 30+ conversions recorded in any rolling 30-day window
Stable / Optimizing	30–100/month	Maximize Conversions + tCPA set close to observed CPA	After 2–4 week stable period at tCPA without delivery throttling
Mature / Value-based	100+/month with reliable values	Maximize Conversion Value + tROAS set 10% below observed ROAS	After 4-week ROAS observation period; adjust tROAS in ≤ 15% increments

Target-setting rules

• Set tCPA close to your maximum acceptable cost per lead or purchase, not an aspirational target. A target that is too low restricts delivery and can push the campaign back into a learning state.^[12]
• Set initial tROAS 10% below observed ROAS to preserve serving volume, then raise in increments of no more than 15% every 14 days.^[10][11]
• Make only one significant change (bid strategy, target value, or budget) at a time. Concurrent changes compound learning-period instability.^[13]
• Budget should allow for up to 2× daily average spend on high-opportunity days. A tightly capped budget will constrain the algorithm independently of the bid target.^[13]

8. Conversion Value Rules

Conversion value rules allow you to adjust the value the Smart Bidding algorithm assigns to a conversion based on characteristics such as device, location, or audience segment. They do not change reported revenue; they change the value signal sent to the bidding model, directing it toward the conversions you actually care about most.^[18]

When to use value rules

• Geographic value differences: If a lead from New York City converts to a closed deal at twice the rate of a lead from a rural market, assign a 2× multiplier to NYC conversions so the algorithm bids proportionally higher for NYC auctions.^[18]
• Device value differences: If desktop purchase orders average $180 and mobile orders average $95, assign device-specific value adjustments so the algorithm does not over-invest in the lower-value channel.^[18]
• Audience value differences: If returning customers have a 3.2× higher LTV than new customers, assign a 3.2× value multiplier to conversions from returning-customer segments so the algorithm prioritizes retention alongside acquisition.^[18]

Implementation rules

• Value rules operate at the campaign level. Set them in the campaign’s conversion settings, not at the asset group level.^[18]
• Combine value rules with tROAS bidding for maximum effect. With tCPA (count-based) bidding, value rules have no mechanism to influence bid levels.^[18]
• Audit value rules quarterly. If underlying LTV or close-rate data changes, outdated multipliers will misdirect the algorithm.^[18]

9. Brand Exclusions and Brand Safety

Brand exclusions prevent PMax from capturing traffic that you are already winning organically or through a dedicated brand Search campaign, and from serving on placements that conflict with your brand standards. Without proactive exclusions, PMax will typically absorb brand-query conversions and report them, inflating apparent performance while cannibalizing cheaper organic clicks.^[8]

Brand exclusions in PMax

• Apply brand exclusions at the campaign level using Google’s brand exclusion tool inside PMax settings. You can specify brand name variants, misspellings, and associated product names.^[8]
• Brand exclusions are not equivalent to negative keywords; they operate on a brand-entity matching system. Review the list of matched brand entities Google returns before confirming the exclusion.^[8]
• If you are running a separate brand Search campaign, brand exclusions in PMax prevent the two campaigns from competing against each other in the same auction, which would raise your own CPCs.^[8]
• First-party audience exclusions are now supported in PMax as of 2026, allowing you to suppress specific customer lists from receiving PMax ads — useful for excluding recent purchasers from acquisition campaigns.^[3]

Placement exclusions and brand safety

• Use the “When and where ads showed” placement report (now segmented by network) to identify low-quality or brand-unsafe placements receiving budget.^[3]
• Apply placement exclusions at the account level for any domain or app category that consistently conflicts with brand guidelines — for example, explicit content, politically partisan content, or competitor-branded properties.^[7]
• Review placement reports monthly during the first 90 days of a new PMax campaign and quarterly thereafter. Budget flowing to irrelevant placements is a signal that audience signals need sharpening, not just that exclusions are needed.^[3]

10. Reporting, Channel Controls, and Transparency

A common criticism of PMax has been its opacity. Google has responded with a sustained rollout of reporting features throughout 2026, including channel-level performance breakdowns, network-segmented placement reports, budget projection tools, and expanded audience demographic reporting. Knowing where to look is now a core PMax management skill.^[3][19]

Key reports and where to find them

• Channel performance report: Shows spend, conversions, and cost metrics broken out by channel (Search, YouTube, Display, Discover, Gmail, Maps). Access via the campaign’s Insights and reporting menu. Use this to assess whether a specific channel is consuming disproportionate budget relative to conversion contribution.^[3][19]
• Placement report (network-segmented): Found under “When and where ads showed.” Now segmented by network, allowing you to see which specific domains, apps, or YouTube channels your Display and Video budget is reaching. Review weekly for the first 60 days of a new campaign.^[3]
• Budget report: Projects end-of-month spend and models the impact of budget changes. Use this before raising or cutting daily budget to understand the projected 30-day effect.^[3]
• Audience report: Now includes demographic breakdowns by age, gender, and audience segment. Use this to validate whether the algorithm is reaching the demographic profile you expected from your audience signals.^[3]
• Insights page: Provides top search themes, audience insights, and asset performance. Consult this before making structural changes and after any significant performance shift.^[8][14]
• Asset performance labels: Best / Good / Low labels at the individual asset level. Use these alongside Ad Strength — do not optimize by Ad Strength alone.^[4][14]

Conversion lag and evaluation discipline

• Always account for conversion lag — the gap between a click and a recorded conversion — before concluding that a budget or bid change improved or hurt performance. For B2B accounts with multi-week sales cycles, do not evaluate a change until at least one full conversion cycle has elapsed.^[8]
• Use the Insights page to check whether recent performance changes are driven by creative, audience, or market factors before adjusting bid strategy.^[14]

11. Common Mistakes to Avoid

The following errors appear repeatedly in PMax accounts and are responsible for the majority of wasted spend and misattributed performance gains. Each is specific and actionable.

Structural and setup mistakes

• Launching without conversion tracking in place. PMax with no reliable conversion data is bidding blind. Verify that tags fire on 100% of conversion events before launch and that conversion actions are set to Primary.^[3][8]
• Adding a tCPA or tROAS before the account has enough volume. Applying a tCPA when the account records fewer than 30 conversions per month over-constrains the algorithm. Start with Maximize Conversions until the threshold is met.^[3][12]
• Running PMax alongside Standard Shopping without a formal Experiment. PMax takes auction priority over Standard Shopping for the same products. Running both wastes budget and makes attribution ambiguous.^[9]
• Omitting video assets. Accounts that launch without video force Google to auto-generate assets or forgo YouTube inventory entirely. Supply at least two videos in 16:9 and 9:16 format at launch.^[4][9]

Bidding and optimization mistakes

• Setting an aspirational tCPA or tROAS far below observed performance. A tCPA set 30% below observed CPA will throttle delivery within days. Start within 10% of observed performance.^[12]
• Making concurrent changes during the learning period. Changing the bid strategy, budget, and creative simultaneously resets the learning period each time. One change at a time, minimum 14 days apart.^[2][13]
• Evaluating performance before conversion lag resolves. A B2B account with a 21-day sales cycle cannot be evaluated 7 days after a change. Allow at least one full conversion cycle before drawing conclusions.^[8]

Asset and creative mistakes

• Uploading repetitive or interchangeable headlines. If three headlines all say variations of “Best [Product] in America,” the algorithm has no meaningful creative variation to test. Each headline should lead with a distinct value proposition.^[5]
• Removing assets too quickly based on Low labels. Google recommends waiting 2–3 weeks for labels to stabilize and always adding new assets before removing underperformers.^[4][14]
• Relying entirely on AI-generated assets. Auto-generated assets reduce friction but produce lower creative quality than purpose-built first-party assets. Treat them as a fallback, not a strategy.^[2][9]

Measurement and brand-protection mistakes

• Not applying brand exclusions before launch. Without brand exclusions, PMax will absorb brand-query traffic and report those conversions as incremental, inflating apparent campaign performance.^[8]
• Never reviewing the placement report. Budget flowing to off-topic apps, low-quality websites, or brand-unsafe domains is invisible unless you check the placement report at least monthly.^[3]
• Treating PMax ROAS as directly comparable to Search ROAS. PMax blends upper-funnel channels (YouTube, Display) with lower-funnel Search. Its blended ROAS will typically be lower than a pure Search ROAS benchmark; comparing them without channel-level data misleads optimization decisions.^[4][17]

12. What Changed Recently (Last 30 Days)

Google has used the September–October 2026 period to expand PMax reporting and steering controls significantly, signaling a continued shift toward greater transparency without changing the core bidding automation model. The changes below are confirmed in Google’s official announcements and API documentation as of October 2026; individual features may still be rolling out and may not yet be visible in every account.^[3][19]

Expanded reporting

• Channel-level performance reporting is now broader. Advertisers can see spend, conversions, and cost-per-conversion segmented by channel (Search, YouTube, Display, Discover, Gmail, Maps) inside PMax. Use this data to validate whether your asset strategy is serving appropriately across surfaces before making structural changes.^[3][19]
• Placement reports now segmented by network. The “When and where ads showed” tab now allows network-level segmentation, making it possible to distinguish Search placements from Display placements in the same report.^[3]
• Budget reports are now available inside PMax campaigns. Advertisers can use the in-campaign budget tool to project end-of-month spend and model the effect of daily budget changes before committing.^[3]
• Expanded audience demographic reporting. Full demographic breakdowns including age and gender are now available inside PMax, allowing advertisers to validate whether signals are steering the algorithm toward the expected buyer profile.^[3]

Audience and targeting controls

• First-party audience exclusions are now supported. You can exclude specific customer lists from receiving PMax ads — a significant addition for accounts that want to prevent retention campaigns from serving acquisition creative to existing customers.^[3]
• Asset experiments are expanding. Google has announced that asset-level experiments will become available in PMax, enabling controlled creative tests without requiring a full campaign experiment.^[19]

Local Services Ads migration

• Local Services Ads are migrating into PMax as pay-per-lead campaigns in phases throughout 2026. Manual bidding is no longer supported in this migration path. If your account includes Local Services Ads, verify your current migration status and ensure your PMax conversion tracking and budget settings are configured for the pay-per-lead model before the migration reaches your account.^[15][16]

Practitioner guidance for the current moment

• Pull the new channel and placement reports before making any structural campaign changes. The expanded visibility means decisions that previously required inference can now be data-driven.^[3][19]
• Use the budget projection tool before raising or cutting daily budget, particularly heading into Q4 2026 seasonal periods where actual spend can approach 2× daily average on peak days.^[3][13]
• If managing service-based businesses, audit any Local Services Ads campaigns immediately for migration impact and confirm that PMax conversion tracking is correctly configured for pay-per-lead attribution.^[15][16]
• Begin planning asset experiments for Q1 2027 creative refresh cycles, now that the experiment framework is expanding into PMax asset-level testing.^[19]

---

References

1. [1] https://business.google.com/us/accelerate/resources/articles/audience-signals-in-performan… business.google.com
2. [2] https://developers.google.com/google-ads/api/performance-max/structure-requests developers.google.com
3. [3] https://developers.google.com/google-ads/api/performance-max/asset-requirements developers.google.com
4. [4] https://support.google.com/google-ads/answer/14528220?hl=en support.google.com
5. [5] https://services.google.com/fh/files/misc/performance_max_best_practices_guide.pdf services.google.com
6. [6] https://www.thinkwithgoogle.com/_qs/documents/18344/Google_UKI___Creative_in_Performance_M… www.thinkwithgoogle.com
7. [7] https://support.google.com/google-ads/answer/17091269?hl=en support.google.com
8. [8] https://support.google.com/google-ads/answer/14753570?hl=en support.google.com
9. [9] https://support.google.com/google-ads/answer/10724492?hl=en support.google.com
10. [10] https://support.google.com/google-ads/answer/14528221?hl=en support.google.com
11. [11] https://developers.google.com/google-ads/api/performance-max/assets developers.google.com
12. [12] https://developers.google.com/google-ads/api/performance-max/retail developers.google.com
13. [13] https://business.google.com/us/accelerate/resources/articles/multiply-conversions-with-per… business.google.com
14. [14] https://support.google.com/google-ads/answer/13872527?hl=en support.google.com
15. [15] https://abc17news.com/stacker-business-economy/2026/09/16/google-local-services-ads-transi… abc17news.com
16. [16] https://ground.news/article/local-services-ads-to-performance-max-what-to-audit-before-mig… ground.news
17. [17] https://developers.google.com/google-ads/api/performance-max/asset-groups developers.google.com
18. [18] https://support.google.com/google-ads/answer/14530785?hl=en support.google.com
19. [19] https://business.google.com/us/accelerate/resources/articles/new-performance-max-steering-… business.google.com