Headlines and Descriptions

Last updated: October 16, 2026

1. Executive Summary

Responsive Search Ads (RSAs) remain Google’s only actively supported Search ad format as of October 2026. The format gives copywriters up to 15 headlines and 4 descriptions; Google’s machine-learning system then assembles and tests combinations automatically. Five principles govern every high-performing RSA:

• Maximize distinct assets, not just asset count. Submit all 15 headlines and all 4 descriptions wherever possible, but make every asset earn its place with a unique message angle. Near-duplicates waste combination slots and can suppress Ad Strength.^[1]
• Lead with relevance, then sell the benefit. Include the core keyword or intent signal in at least one to two headlines so the ad feels immediately relevant, then use the remaining headlines to communicate outcomes, proof, and differentiation that no keyword alone can convey.^[2]
• Write every asset to stand alone. Because Google can display any combination of up to three headlines and two descriptions, each headline and each description must make sense without relying on another asset appearing beside it.^[1]
• Treat Ad Strength as a setup diagnostic, not a performance guarantee. Research on approximately 20,000 Google Ads accounts found that Average Ad Strength outperformed Excellent on conversion rate (12.65% vs. 4.97%) and CPA ($12.43 vs. $28.68 CPA), which directly contradicts the vendor claim that Excellent always wins.^[3] Use Ad Strength to fix coverage gaps, but judge creative quality by conversion data.
• Pair human-authored copy with controlled AI generation. As of September 2026, Google is migrating automatically created assets (ACA) into AI Max for Search as “text customization.” Supply a complete, high-quality human-written RSA first, then use text-guideline controls and asset previews to govern what AI generates alongside it.^[4]

2. Benchmarks and Numbers at a Glance

Metric	Typical range or threshold	Applies when	Source
Maximum headlines per RSA	15 (minimum 3)	All RSAs; Google-enforced hard limit	[1]
Maximum descriptions per RSA	4 (minimum 2)	All RSAs; Google-enforced hard limit	[1]
Headline character limit	30 characters including spaces and punctuation	All RSAs; double-width CJK characters count as 2	[1]
Description character limit	90 characters including spaces and dynamic text	All RSAs	[1]
Ad Strength: Poor → Excellent incremental lift	+15% clicks and conversions on average (vendor claim, sample size not stated)	Vendor claim from Google; not a guaranteed per-account result	[5]
Adding a 2nd RSA per ad group	+6.6% conversions at similar CPA (vendor claim, sample size not stated)	Vendor claim from Google; assumes Good or Excellent Ad Strength on both RSAs	[5]
Adding a 3rd RSA per ad group	+3.7% conversions at similar CPA (vendor claim, sample size not stated)	Vendor claim from Google; diminishing returns above 3 RSAs are typical	[5]
Description length sweet spot (CTR + CVR)	61–70 characters: 12.33% CTR, 9.21% CVR, 307.58% ROAS	Study across ~20,000 Google Ads accounts; correlational, not causal	[3]
Short descriptions (0–50 characters)	8.90% CTR, $9.61 CPA	Study across ~20,000 Google Ads accounts; shorter copy correlates with lower CTR	[3]
Average Ad Strength conversion rate	12.65% CVR, $12.43 CPA	Study across ~20,000 accounts; contradicts vendor claim that Excellent always outperforms	[3]
Excellent Ad Strength conversion rate	4.97% CVR, $28.68 CPA	Study across ~20,000 accounts; note higher CPA than Average Ad Strength	[3]
Partial pinning performance	11.88% CTR, 10.55% CVR, $13.68 CPA, 365.15% ROAS	Study across ~20,000 accounts; partial pinning outperformed full pinning on all four metrics	[3]
Full pinning performance	8.49% CTR, $32.57 CPA	Study across ~20,000 accounts; full pinning correlated with meaningfully worse efficiency	[3]
Expanding from 3 to 14 headlines	CTR increase from 3.2% to 4.9% (+53%) (vendor claim, sample size not stated)	Vendor claim; directional evidence for filling all headline slots with distinct copy	[6]
Ad Strength: higher-CTR group	11.77% CTR, 10.39% CVR	Study across ~20,000 accounts; context-dependent; do not treat as a universal benchmark	[3]

Note on conflicting data: Google’s vendor claim states that Excellent Ad Strength produces 15% more clicks and conversions than Poor Ad Strength.^[5] The independent Optmyzr study of approximately 20,000 accounts found that Average Ad Strength delivered a higher conversion rate (12.65%) and lower CPA ($12.43) than Excellent Ad Strength (4.97% CVR, $28.68 CPA).^[3] These findings conflict directly. The most conservative interpretation: improve Ad Strength to eliminate obvious coverage gaps, but do not inflate asset count with weak copy simply to reach Excellent. Judge outcomes by conversion data.

3. How Responsive Search Ads Assemble Your Headlines and Descriptions

A Responsive Search Ad is a multivariate creative system, not a static ad unit. When you submit up to 15 headlines and 4 descriptions, Google does not show them all at once. Instead, its auction-time system selects and arranges assets to fit the available ad slot, the device, the query, and the predicted user intent.^[1]

What Google shows

• Headlines: Up to 3 headlines appear in a single impression, separated by vertical pipes (|). Google may show 1, 2, or 3 headlines depending on available space and device constraints.^[1]
• Descriptions: Up to 2 descriptions appear per impression. Fewer may display on mobile or in compressed ad formats.^[1]
• Order: Headlines can appear in any order unless you pin them to a specific position. The system tests different sequences to find combinations that generate the most relevant, high-performing impressions for each query context.^[7]

Combination math

With 15 headlines and 4 descriptions, the theoretical number of ordered 3-headline combinations exceeds 2,700 before accounting for description pairings. Google does not test every combination with equal frequency; its machine-learning system allocates more impressions to combinations predicted to perform well for the query and audience.^[7]

Practical implications for writers

• Every headline must be coherent in position 1, position 2, or position 3 without depending on which other headlines accompany it.
• Avoid headlines that are sentence continuations of each other (e.g., “Get Fast” and “Reliable Service” only work together).
• Descriptions should be complete, standalone messages—not fragments that need a specific headline to make sense.
• Check problematic headline combinations manually. A headline saying “Lowest Prices” alongside one saying “Premium Quality Service” may appear contradictory to a user who sees both simultaneously.
• Asset ratings in the Google Ads interface reflect relative performance across the combinations Google tested, not the quality of an asset in every possible context.^[8]

How the system learns

The RSA system requires conversion volume to learn effectively. Google does not publish an official minimum threshold, but the general principle in performance bidding applies: more conversion data per period enables faster, more reliable signal. Accounts with low conversion volume will see slower asset differentiation and less reliable asset ratings. This is an additional reason to launch with a complete, high-quality set of 15 headlines and 4 descriptions from the start rather than adding assets incrementally after launch.^[1]

4. Writing Effective Headlines

Headlines are the most visible element of a Search ad. They carry the keyword relevance signal, the value proposition, and the first impression of brand credibility—all within a 30-character limit per asset. Because Google displays up to three headlines per impression in any order, each headline must function independently while contributing to a coherent overall message when combined with others.^[1]

The 30-character constraint in practice

Thirty characters is approximately five to seven short words. Every character is meaningful. Use contractions where they sound natural (“You’re,” “We’ll”), abbreviations where they are universally understood (“CPA,” “HVAC,” “SaaS”), and numerals instead of spelled-out numbers (“20%” not “twenty percent”). Punctuation counts toward the limit; em dashes, asterisks, and trademark symbols consume characters and should be used only when they add genuine value.

A recommended 15-headline composition

Slot range	Headline role	Purpose	Example (≤30 chars)
1–2	Keyword-focused	Match search intent; signal relevance	“Emergency Plumber Denver”
3	Brand + category	Anchor brand identity	“Acme Plumbing & Drain”
4–5	Primary benefit	Answer “what do I gain?”	“Stop Leaks Fast—Same Day”
6–7	Secondary benefit	Address a second decision factor	“Upfront Pricing, No Surprises”
8	Problem solved	Speak to the pain driving the search	“Burst Pipe? We Fix It Today”
9	Differentiator	Explain why this provider vs. others	“Licensed & Insured Plumbers”
10	Social proof	Reduce perceived risk with evidence	“4.9 Stars · 1,200 Reviews”
11	Guarantee / reassurance	Remove objection or fear	“Satisfaction Guaranteed”
12	Price / starting price	Qualify intent; filter bargain vs. budget	“Service Calls From $89”
13	Promotion or offer	Create urgency with a real incentive	“$30 Off First Repair—June”
14	Specific CTA	Direct the next action clearly	“Book A Same-Day Visit Now”
15	Brand CTA or urgency	Reinforce brand and close	“Call Acme—Available 24/7”

Leading with the keyword vs. leading with the benefit

Lead with the keyword when the searcher is in transactional mode and category match is the primary buying signal: “Business Tax Software,” “Running Shoes Online,” “HVAC Repair Near Me.” Lead with the benefit when the outcome is more persuasive than the category label: “Cut Energy Bills 20%,” “File Taxes In Under 10 Min,” “Shoes Built For Long Runs.” The strongest RSA portfolio includes both angles across the 15 assets.^[2]

Numbers, prices, and specificity

Specific numbers make claims scannable and concrete. “Save 20% Today,” “Plans From $29/Mo,” “4.8-Star Rated Service,” and “Free 14-Day Trial” all communicate a verifiable advantage. Avoid unsupported superlatives: “#1,” “Best Price,” “Lowest Rates,” and “Guaranteed Savings” are either unverifiable or banned under Google Ads policy unless you can substantiate them on the landing page.^[2]

Calls to action in headlines

Include at least one headline with a specific CTA that matches the landing-page action. Avoid generic CTAs in headline slots: “Click Here,” “Learn More,” and “Buy Now” consistently underperform more specific alternatives. Stronger options include “Get Your Free Quote,” “Book A Same-Day Visit,” “Start Your Free Trial,” and “Compare Plans Online,” each of which tells the user precisely what happens after the click.^[2]

Headline variety: angle, not just wording

Variety means changing the message angle—not shuffling the same claim into different word orders. A set of 15 headlines that all say “Fast Service” in slightly different ways gives Google almost no meaningful combinations to test. Aim to cover at least eight to ten distinct message angles from the list: search intent, primary benefit, secondary benefit, product feature, brand, social proof, guarantee, price, promotion, delivery or convenience, specific CTA, objection handling, and audience or location relevance.^[2]

5. Writing Effective Descriptions

Descriptions carry the load of persuasion that headlines cannot accomplish in 30 characters. Each description has a 90-character limit and serves as a standalone selling argument. Google may display up to two descriptions per impression, in any pairing, alongside any eligible headline combination—so every description must remain accurate and compelling regardless of which headlines appear above it.^[1]

Use all four description slots

Providing only two descriptions is the minimum; providing all four gives Google twice the combinations to test and ensures broader coverage of your selling argument. The four-slot framework below distributes the persuasive work across independent, non-duplicative messages:^[1]

Slot	Primary job	What to include	Example (≤90 chars)
1	Core value proposition	Product or service + main customer benefit	“Expert small business tax filing by licensed CPAs—accurate returns, maximum refund.”
2	Differentiation or mechanism	Feature, process, speed, convenience, or unique advantage	“Upload documents securely online. Our team reviews and files—no office visit needed.”
3	Social proof	Customer count, rating, award, certification, years in business	“Trusted by 8,000+ businesses nationwide. Rated 4.9/5 by verified clients.”
4	Offer and action	Discount, free trial, free quote, deadline, specific CTA	“Book your free tax consultation today—no obligation, no hidden fees.”

Description length and performance

Research across approximately 20,000 Google Ads accounts found that descriptions in the 61–70 character range correlated with the strongest combined performance: 12.33% CTR, 9.21% conversion rate, and 307.58% ROAS. Descriptions of 0–50 characters correlated with materially lower performance: 8.90% CTR and $9.61 CPA.^[3] The implication is not that you should pad descriptions to hit 65 characters, but that descriptions significantly shorter than 70 characters likely leave persuasive capacity unused. Write until you have said the most compelling version of your message, then count characters.

Calls to action in descriptions

Reserve at least one of your four descriptions for a strong, specific CTA matched to the landing-page action. Avoid repeating the same CTA across multiple descriptions; vary the angle to test different user motivations:^[2]

• Transactional: “Order Online Today,” “Claim 15% Off Your First Purchase”
• Lead generation: “Request A Free Quote—Response Within 2 Hours”
• Evaluation: “Compare Plans And Pricing—No Signup Required”
• Software / trial: “Start Your 14-Day Free Trial—No Credit Card Needed”
• Informational: “Download The 2026 Buyer’s Guide—Free”

Social proof standards

Social proof works best when it is specific, verifiable, and tied to a real number or credential. “Trusted By Many” is weaker than “Trusted By 10,000+ Small Businesses.” “Great Reviews” is weaker than “Rated 4.8/5 By 2,000+ Verified Customers.” Do not claim a rating, award, or customer count you cannot substantiate on the landing page.^[2]

Consistency with headlines and landing page

Every description must remain accurate alongside any eligible headline combination. If a headline says “Plans From $29/Mo,” no description should suggest that all plans are free. If a description says “Sale Ends This Sunday,” the landing page must show the same end date. Mismatches between ad copy and landing-page content violate Google’s advertising policies and erode user trust.^[2]

6. Copywriting Frameworks and When to Use Them

No single framework produces the best RSA in every context. The right structure depends on the industry, product type, funnel stage, and season. The guidance below explains each framework, when to deploy it, and how to adapt it for the RSA format’s asset-by-asset construction. Use the comparison table to select your primary framework, then build a modular RSA that borrows elements from others as needed.^[10]

Framework comparison

Framework	Best For (industry / product)	Funnel Stage	Example Headline or CTA
FAB (Features–Advantages–Benefits)	SaaS, e-commerce, technology, appliances, professional services	Consideration → Conversion (mid-to-bottom funnel)	“Automate Follow-Ups” → “Close More Deals Faster”
PAS (Pain–Agitate–Solution)	Home services, legal, healthcare, cybersecurity, financial services	Problem-aware → High-intent (bottom funnel)	“Burst Pipe? Stop Water Damage Fast” → “Book A Same-Day Visit”
AIDA (Attention–Interest–Desire–Action)	B2B software, education, real estate, travel, unfamiliar categories	Awareness → Consideration (top-to-mid funnel)	“Learn Data Analytics Online” → “Enroll For The Next Cohort”
4 Ps (Promise–Picture–Proof–Push)	Legal, healthcare, agencies, premium e-commerce, B2B services	Consideration → Conversion (mid-to-bottom funnel)	“Cut Monthly Energy Costs” → “Claim Your Free Quote Today”
Storytelling (micro-narrative)	Travel, lifestyle, food, apparel, fitness, nonprofit, consumer brands	Awareness → Remarketing (top funnel and re-engagement)	“Your Next Adventure Starts Here” → “Find Your Perfect Trail Gear”

1. Features–Advantages–Benefits (FAB)

What it is: FAB translates a product specification (Feature) into how that specification performs (Advantage), then into what that means for the customer (Benefit). In a 30-character headline environment, benefits should receive the most emphasis because Google’s guidance explicitly recommends messaging that focuses on user needs and outcomes.^[10]

Why it works: FAB answers the three questions behind a click—”What is it?”, “How is it different?”, and “What will I gain?”—in a structured sequence that gives Google independent assets to assemble: a feature headline, a benefit headline, a proof description, and a CTA description.

When to use it: SaaS, e-commerce, software, appliances, automotive, and professional services. Strongest at the consideration and conversion stages where users are comparing alternatives. Use evergreen FAB copy as the base, then adapt individual benefit headlines to seasonal priorities: “Fast Holiday Delivery” in Q4, “Tax-Season Bookkeeping” in February–April, “Back-to-School Savings” in July–August.

CTA style: Action-oriented and outcome-specific. “Start Your Free Trial,” “Get A Live Demo,” “See Pricing Now,” “Download The Spec Sheet.”

RSA example (CRM software):

• Headline: “CRM Software For Sales Teams” (feature / category)
• Headline: “Automate Lead Follow-Ups” (advantage)
• Headline: “Close More Deals Faster” (benefit)
• Description: “Centralize contacts, automate outreach and track every deal in one dashboard. Try free for 14 days.”

2. Pain–Agitate–Solution (PAS)

What it is: PAS identifies a recognized customer problem (Pain), intensifies its consequences to create urgency (Agitate), and presents the product as the credible resolution (Solution). In search advertising, agitation must be factual and proportionate—not sensational or fear-based.

Why it works: PAS aligns with problem-aware searches where the user has already identified the issue. Someone searching “ransomware protection small business” or “IRS tax debt help” needs acknowledgment of the problem before they will respond to a benefit claim. Google’s own guidance warns against generic messaging and favors specific, intent-aligned copy—PAS delivers this naturally.^[10]

When to use it: Emergency home services, legal services, healthcare, cybersecurity, debt relief, IT support, and insurance. Bottom-funnel, high-intent searches. Seasonally strongest during storm damage periods, tax deadlines (January 15–April 15), open enrollment windows (October 15–December 7 for Medicare, November–December for ACA), and flu season. Do not apply artificial urgency when no genuine deadline exists.

CTA style: Relief-oriented and immediate. “Get Help Now,” “Call For Emergency Service,” “Book A Same-Day Visit,” “Talk To A Tax Expert Today.”

RSA example (roofing emergency):

• Headline: “Leaking Roof? Get Help Now” (pain)
• Headline: “Stop Water Damage Fast” (agitate)
• Headline: “Free Roof Inspection Today” (solution)
• Description: “A leaking roof can cause $10,000+ in interior damage within 48 hours. Call local specialists now.”

3. AIDA (Attention–Interest–Desire–Action)

What it is: AIDA is a four-stage persuasion sequence. In an RSA, it functions better as a portfolio of assets than a linear single ad—keyword-aligned headlines earn attention, feature and differentiator headlines create interest, benefit and proof assets build desire, and specific CTA assets drive action.^[7]

Why it works: AIDA maps naturally onto RSA asset roles. Google’s system can assemble attention-stage and desire-stage assets in the same impression, or serve a desire + action pairing to a user who already knows the category. This flexibility makes AIDA highly compatible with RSA’s multivariate format.

When to use it: B2B software, education, financial products, real estate, travel, and any category where users need explanation or trust before converting. Useful across upper-to-middle funnel. Seasonally strongest for product launches, annual enrollment periods (October–December), and event promotions.

CTA style: Consideration-stage CTAs work well: “Compare Options,” “See How It Works,” “Enroll Today,” “Request A Callback,” “Download The Guide.”

RSA example (online analytics course):

• Headline: “Learn Data Analytics Online” (attention)
• Headline: “Flexible Expert-Led Courses” (interest)
• Headline: “Build Job-Ready Skills Fast” (desire)
• Description: “Master analytics through practical projects, mentor support and flexible lessons. Apply for the next cohort now.”

4. The 4 Ps (Promise–Picture–Proof–Push)

What it is: The 4 Ps framework structures persuasion around a clear Promise of the primary result, a Picture of the improved situation (in ad copy, a concise benefit or outcome statement), Proof of credibility (evidence, ratings, certifications), and a Push—a reason to act now. Google specifically recommends including benefits, promotions, trust signals, and clear next actions in RSA messaging, making 4 Ps a strong structural fit.^[10]

Why it works: The framework balances motivation (Promise + Picture) with risk reduction (Proof) and urgency (Push). It is particularly effective where customers hesitate because of price, trust, complexity, or perceived risk—common in high-ticket or professional services.

When to use it: Legal, healthcare, financial services, premium e-commerce, home improvement, and B2B. Consideration and conversion stages. Seasonally strongest during promotional periods, renewal windows, Black Friday (late November), tax deadlines, and limited-slot service availability campaigns.

CTA style: Risk-reduction and incentive-oriented. “Claim Your Free Quote,” “Get A No-Obligation Estimate,” “See If You Qualify,” “Book Your Spot Before It Fills.”

RSA example (home insulation):

• Headline: “Cut Monthly Energy Bills” (promise)
• Headline: “Enjoy A More Efficient Home” (picture)
• Headline: “4.9-Star Local Installers” (proof)
• Description: “Upgrade to certified insulation with transparent pricing and trained local crews. Get a free estimate today.”

5. Storytelling (micro-narrative)

What it is: Storytelling in a Search ad compresses a narrative arc—a situation, a transformation, and a better outcome—into a single headline-and-description pair. At 30 characters per headline, there is no room for a full story; instead, use a customer moment, an aspirational image, or a relatable situation to create emotional resonance.

Why it works: Stories make otherwise interchangeable benefits concrete and memorable. They are most valuable when the purchase involves identity, aspiration, or a meaningful life event—not merely a functional transaction. Storytelling should still connect directly to search intent; a vivid message that does not answer the query weakens relevance and landing-page continuity.^[10]

When to use it: Travel, hospitality, food, apparel, fitness, nonprofit, consumer packaged goods, and lifestyle brands. Strongest at awareness stage and in remarketing campaigns. Seasonally excellent around major gifting moments: Valentine’s Day (February 14, 2027), Mother’s Day (May 10, 2026), Thanksgiving–Cyber Monday (November 26–30, 2026), and Christmas (December 2026).

CTA style: Experiential and exploratory. “Find Your Next Adventure,” “Give A Gift They’ll Remember,” “Start Your Journey Today,” “Discover What’s Waiting.”

RSA example (travel gear, holiday remarketing):

• Headline: “Every Trail Tells A Story” (situation)
• Headline: “Gear Built For The Journey” (transformation)
• Headline: “Shop Holiday Gifts—Free Ship” (action + offer)
• Description: “Give the outdoor adventurer in your life gear built to last. Free shipping on all orders through December 31, 2026.”

7. Pinning Strategy

Pinning locks a specific headline or description to a specific position—Headline 1, Headline 2, Headline 3, or Description 1–2—so Google always serves that asset in that slot. It is the only mechanism in RSAs that guarantees a message appears in a defined position.^[1] Used surgically, pinning protects critical messaging. Used carelessly, it eliminates most of the combinatorial testing that makes RSAs effective.

When pinning is justified

• Legal or regulatory requirements: Disclaimers mandated by law, financial disclosures, healthcare compliance language, or state-specific restrictions that must appear in every impression.
• Brand standards: A franchisor or legal department that requires the brand name in a specific position for every ad in a specific campaign.
• Contractual guarantees: A promotion with partner-required terms that must be visible to qualify for co-op funding.
• Structural coherence: A two-part message where Headline 2 only makes sense after a specific Headline 1—for example, “Zero-Fee Balance Transfers” pinned to H1, “For 15 Months” pinned to H2.

When pinning is not justified

• You prefer a particular headline and want it to show most of the time.
• You want to approximate the experience of an expanded text ad.
• You are pinning to “control” the ad and feel uncomfortable with Google’s automatic combinations.
• You are pinning every asset because Ad Strength dropped when you left assets unpinned.

The performance cost of over-pinning

A study across approximately 20,000 Google Ads accounts found that fully pinned RSAs produced an 8.49% CTR and a $32.57 CPA, while partially pinned RSAs produced an 11.88% CTR, a 10.55% conversion rate, and a $13.68 CPA.^[3] A separate study across more than 1 million ads found that ads with some or no pinning outperformed fully pinned ads on conversion rate and cost-based metrics, while fully pinned ads showed stronger relevance-based metrics such as CTR in certain segments.^[11] These two findings point in slightly different directions on CTR; the most conservative interpretation is that partial pinning—where only operationally essential slots are locked—tends to preserve conversion performance better than full pinning, while also avoiding the relevance loss associated with zero pinning in highly competitive queries.

How to pin without destroying coverage

• Pin multiple assets to the same position. Most campaign interfaces allow you to assign two or three headlines to Headline 1. Google then rotates among the pinned options, restoring some variety while keeping the position protected.
• Pin only the position that is operationally required. If a disclaimer must appear somewhere in the ad, pin it to Description 2 and leave H1, H2, H3, and D1 free.
• Do not pin a CTA. Google’s system already optimizes CTA placement; pinning a generic CTA to H3 wastes a slot that could carry a benefit or proof point.
• Document the reason. Note in campaign comments why each pin exists. If the reason no longer applies—the promotion ended, the legal requirement was lifted—remove the pin immediately.

Pinning and Ad Strength

Google’s Ad Strength evaluator penalizes heavy pinning because it reduces the number of valid combinations. If Ad Strength drops after pinning, that reflects reduced combinatorial coverage—not necessarily reduced ad quality. Do not unpin compliance-required assets to chase an Excellent rating. Instead, offset the coverage loss by adding more unpinned headlines and descriptions in the remaining slots.^[8]

Pinning level	CTR (study)	Conversion rate (study)	CPA (study)	Recommendation
No pinning	Comparable to partial pinning	Strong	Low	Default for most RSAs[11]
Partial pinning (1–2 positions)	11.88%	10.55%	$13.68	Use only when operationally required[3]
Full pinning (all positions)	8.49%	Lower than partial	$32.57	Avoid; use only under strict mandate[3]

8. Keyword Insertion and Dynamic Text Customization

Google Ads offers two primary mechanisms for inserting dynamic text into RSA copy: keyword insertion (also called dynamic keyword insertion, or DKI) and ad customizers. As of October 2026, a third layer—AI-generated text customization within AI Max for Search—also generates dynamic copy from landing-page and domain content, but that feature operates at the campaign level rather than at the individual asset level.^[4]

Dynamic keyword insertion (DKI)

DKI replaces a placeholder in a headline or description with the search query that triggered the ad, up to the character limit. The syntax is {KeyWord:Default Text}, where the default text appears when the query exceeds the character limit or would violate policy.^[12]

When DKI helps:

• Ad groups with tightly themed keyword lists where the query itself is a meaningful headline.
• High-intent, transactional queries where exact-phrase matching adds relevance.
• Large product-catalog campaigns where writing individual headlines for every SKU is impractical.

When DKI hurts:

• Broad match campaigns where the inserted query may be irrelevant, misspelled, or grammatically broken.
• Premium or brand-sensitive contexts where inserting a competitor’s name or a generic query damages positioning.
• RSAs with loosely themed ad groups where the inserted term may contradict other headlines.

DKI does not guarantee better performance. If the default text is weak or the keyword list is broad, DKI can generate awkward, non-persuasive headlines. Always set a strong, specific default such as “Custom Office Furniture” rather than a generic fallback like “Our Products.”

Ad customizers

Ad customizers insert dynamic values—prices, countdown timers, inventory counts, or audience-specific text—from a data feed or a countdown function. Common use cases include:

• Countdown customizers: “Sale Ends in {=COUNTDOWN(“2026-12-01 00:00:00″,”en-US”,2)}” to display a live countdown to a promotional deadline.
• Price customizers: Pull current pricing from a feed so that “From $X/Mo” reflects actual pricing without manual updates.
• Location customizers: Insert a city or region name dynamically to localize copy at scale.

Ad customizers require careful feed management. A stale feed that shows an expired price or an incorrect countdown creates a misleading impression and can violate Google’s advertising policies.^[2]

AI Max text customization

As of September 2026, Google’s automatically created assets (ACA) have been folded into AI Max for Search as text customization. When enabled, the system generates additional headlines and descriptions from the advertiser’s domain, existing ads, and landing-page content. Generated assets are refreshed at least every 48 hours.^[4] Key operational rules:

• Review AI-generated assets using the new preview function before activation. Previews of up to 10 sample headlines and descriptions are available in under 30 seconds.^[13]
• Use text guidelines (term exclusions, messaging restrictions) to prevent prohibited terms or off-brand language from appearing in generated copy.^[4]
• Monitor the asset details report after activation to catch inaccurate claims or policy violations.^[4]
• Do not rely on generated assets as a substitute for human-authored RSA copy. Google says customized text serves only when predicted to outperform advertiser-provided assets.^[4]

Landing-page alignment for dynamic content

Because AI Max text customization draws from landing-page content, keep page titles, meta descriptions, product claims, and pricing current and accurate. A landing page that lists a $49/month price while the RSA shows “$39/month” (whether inserted via DKI, a customizer, or a generated asset) creates a policy risk and a poor user experience.^[2]

9. Ad Strength and Asset Performance Ratings

Ad Strength is Google’s forward-looking diagnostic score for RSAs. It reflects how well the submitted asset set covers relevant messages, avoids duplication, and includes keywords—not how well the ad converts.^[8] Treat it as a setup checklist, not a performance KPI.

The Ad Strength scale

Google’s Ad Strength labels run from Incomplete (missing required fields) through Poor, Average, and Good, to Excellent. Google recommends achieving at least Good for every active RSA.^[8]

Moving an RSA from Poor to Excellent is associated with 15% more clicks and conversions on average, according to Google’s own analysis—but this is a vendor claim without a stated sample size, and the research shows a more nuanced picture.^[5] An independent study across approximately 20,000 accounts found that RSAs with Excellent Ad Strength had a 4.97% conversion rate and a $28.68 CPA, while RSAs rated Average had a 12.65% conversion rate and a $12.43 CPA.^[3] This apparent paradox reflects the fact that highly optimized, tightly controlled RSAs (often with more pinning and fewer combinations) can achieve lower CPAs at the cost of a lower formal Ad Strength score. Do not sacrifice message precision to maximize Ad Strength.

What drives Ad Strength higher

• Submitting 15 distinct headlines (not near-duplicates).
• Submitting all 4 descriptions with meaningfully different content.
• Including the ad group’s primary keyword in at least one headline.
• Avoiding repetition of the same word, phrase, or claim across multiple assets.
• Maintaining variety across message angles: feature, benefit, proof, offer, CTA.
• When text customization is enabled under AI Max, Google-generated assets also contribute to Ad Strength.^[4]

Asset performance ratings

At the individual asset level, Google labels assets Low, Good, or Best based on relative performance in serving contexts.^[8] Interpret these labels as follows:

Rating	Meaning	Minimum data threshold before acting	Recommended action
Best	Outperforms the majority of assets in similar serving contexts	N/A—retain	Keep as a control; develop adjacent variations rather than deleting[8]
Good	Contributing acceptably; not a replacement priority	N/A—monitor	Maintain; test a stronger challenger alongside it
Low	Underperforming relative to other assets, or has insufficient impression data	Establish account-specific threshold (e.g., 500+ impressions) before replacing	Rewrite substantially; check query relevance and landing-page alignment[8]

Asset ratings are relative and contextual. A “Low” headline may underperform because it serves a narrow but high-value query segment, not because the message is objectively weak. Similarly, a “Best” headline may rank highly because it matches the most common query in the ad group, not because it converts the best paying customers. Always cross-reference asset ratings with conversion data and revenue metrics before making replacement decisions.^[8]

Practical Ad Strength workflow

• At launch: Build to Good or Excellent using 15 distinct headlines and 4 descriptions. Fix Incomplete or Poor immediately.
• At 30 days: Review asset performance ratings. Identify any assets rated Low with 500+ impressions. Draft replacement hypotheses.
• At 60 days: Replace 1–3 Low-rated assets with new variants. Do not replace all assets simultaneously, since that destroys the control baseline.
• Ongoing: Keep at least one “Best”-rated asset stable as a control while testing new copy.

10. Testing and Iteration

An RSA is inherently a multivariate testing environment: Google serves different headline and description combinations and accumulates signal about which combinations work best in which contexts.^[14] Effective iteration treats that system as a structured experiment rather than a black box.

Principles of RSA testing

• Test hypotheses, not punctuation. A hypothesis is a meaningful difference in message angle: benefit-led vs. feature-led, price-forward vs. value-forward, urgency vs. trust. Changing a comma to a dash is not a test.
• Change one variable at a time where possible. Replacing 5 headlines and the landing page simultaneously makes it impossible to attribute the result to any single change.
• Use separate RSAs for materially different strategies. If you want to test “lowest price” positioning against “specialist expertise” positioning, those belong in separate RSAs, not mixed into the same asset pool.
• Preserve a stable control. Keep at least one Best-rated asset unchanged while testing new variants. If the new variant underperforms, you have a baseline to return to.
• Evaluate business outcomes, not proxy metrics. Judge RSA performance by qualified conversions, cost per acquisition, revenue, or downstream lead quality—not impressions, Ad Strength, or click-through rate alone.^[8]

How many RSAs per ad group?

Google recommends a minimum of 2 RSAs per ad group, each with Good or Excellent Ad Strength.^[1] Some official guidance references 2–3 RSAs per ad group.^[15] The research benchmark shows that adding a second RSA to an ad group is associated with a 6.6% increase in conversions at a similar cost per conversion, while adding a third is associated with an additional 3.7% increase.^[5] These are vendor claims without stated sample sizes; treat them as directional rather than guaranteed. Running more than 3 RSAs per ad group is rarely justified unless the ad group serves genuinely different audiences, offers, or landing pages with sufficient volume to evaluate each.

Iteration cadence

Review period	What to check	Action threshold	Typical action
7–14 days post-launch	Ad Strength rating, policy approval status, impression delivery	Incomplete or Poor Ad Strength	Add missing headlines or descriptions; fix policy flags
30 days	Asset performance ratings, CTR by device, search term report	Any asset rated Low with 500+ impressions	Draft 1–3 replacement hypotheses; do not act yet
60 days	Conversion volume, CPA, conversion rate by RSA	Statistically meaningful difference between RSAs (account-specific threshold)	Replace 1–3 Low-rated assets; pause underperforming RSA if CPA differs by >25%
90 days	Full creative audit: coverage of all message angles, offer currency, brand alignment	Any asset unchanged for 90+ days with Low rating	Full asset refresh for that RSA; retain Best assets

Using Google’s Campaign Experiments for RSA copy tests

For controlled copy tests where you want to isolate the effect of a messaging change, use Google’s Campaign Experiments tool to split traffic between a base campaign and an experiment campaign. This removes the confound of Google’s internal RSA combination selection and gives you cleaner conversion data. Run the experiment for at least 4 weeks or until each arm has accumulated at least 100 conversions, whichever is longer, before drawing conclusions.

11. Automatically Created and AI-Generated Assets

As of September 2026, Google has begun migrating campaigns that use automatically created assets (ACA) into AI Max for Search, where the equivalent feature is called text customization.^[4] Understanding how this system works—and how to govern it—is now a core RSA operational skill.

How text customization works

When text customization is enabled, Google generates additional headlines and descriptions by analyzing the advertiser’s domain, existing ad copy, landing-page content, and ad-group keywords. Generated assets are:

• Refreshed at least every 48 hours, including when landing-page content changes.^[4]
• Eligible to serve only when Google predicts they will outperform the advertiser-provided assets in a given auction.^[4]
• Counted alongside advertiser-created assets in Ad Strength calculations when the feature is active.^[4]
• Previewable: up to 10 sample generated headlines and descriptions are available in the interface before opt-in, with previews generating in under 30 seconds.^[13]

Governance controls

Google’s new text guidelines allow advertisers to:

• Exclude terms: Prevent specific words or phrases from appearing in generated copy (programmatically supported via term_exclusions in the Ads API).^[4]
• Apply messaging restrictions: Restrict generated copy from referencing certain product categories, claims, or concepts that conflict with brand policy or legal requirements.^[4]
• Monitor via the asset details report: Google specifically recommends checking generated assets for accuracy and policy compliance after activation.^[4]

What you should still write yourself

Text customization does not replace human-authored RSA copy. Google’s system uses your submitted headlines and descriptions as the primary asset pool; generated assets fill gaps or test variants, not lead the creative strategy. A weak human-authored RSA that relies on AI generation for coverage is structurally fragile: if the landing page changes, goes offline, or contains inaccurate pricing, the generated assets may reflect those errors within 48 hours.^[4]

The recommended operating model is:

1. Build a complete, high-quality human-authored RSA (15 headlines, 4 descriptions) covering all key message angles.
2. Ensure the landing page is accurate, up to date, and contains the claims you want reflected in generated copy.
3. Enable text customization via AI Max after reviewing the asset preview.
4. Set term exclusions for any prohibited terminology before launch.
5. Check the asset details report within 7 days of activation and monthly thereafter.

Outside the US

AI Max for Search and text customization availability, rollout timing, and language support vary by market; confirm current availability in the Google Ads interface for non-US accounts before implementing.

12. Common Mistakes to Avoid

The following mistakes consistently degrade RSA performance, increase wasted spend, or create compliance risk. Each is actionable and avoidable.

1. Submitting fewer than 10 distinct headlines

Accounts with only 3–5 headlines leave Google with very few combinations to test, reducing the system’s ability to match copy to query intent. Research shows a progression from 3 headlines at 3.2% CTR to 14 headlines at 4.9% CTR—a 53% increase—though this is a vendor claim without a stated sample size.^[6] The minimum standard is 10 distinct headlines; the target is 15.

2. Near-duplicate headlines that waste asset slots

Submitting “Free Shipping On All Orders,” “We Offer Free Shipping,” and “Get Free Shipping Today” as three separate headlines wastes three slots on a single message angle. Each asset slot should cover a different message angle: the shipping point requires only one headline, freeing the other two slots for proof, pricing, or a CTA.

3. Over-pinning to replicate expanded text ads

Full pinning reduces CTR to 8.49% and raises CPA to $32.57 in available benchmark data, compared to 11.88% CTR and $13.68 CPA for partial pinning.^[3] Pinning every asset to recreate a fixed ad defeats the combinatorial purpose of RSAs. Pin only when an operational, legal, or brand requirement specifically demands it.

4. Writing descriptions as headline extensions

Descriptions that begin with “…and also” or that only make sense after reading a specific headline will fail when Google combines them with different headlines. Every description must be independently coherent as a standalone 90-character message.^[2]

5. Using vague or generic calls to action

“Click Here,” “Learn More,” and “Buy Now” do not tell the user what they will find after the click. Replace generic CTAs with action-specific language: “Request Your Free Quote,” “Book A Same-Day Visit,” or “Download The 2026 Buyer’s Guide.”^[2]

6. Ignoring the landing page

Ad copy that promises “20% Off All Orders Through October 31, 2026” when the landing page shows full-price products creates a policy violation risk and drives immediate bounces. Every claim in the RSA—price, promotion, availability, rating—must be clearly visible and accurate on the landing page.^[2]

7. Treating Ad Strength as a performance KPI

Excellent Ad Strength RSAs showed a 4.97% conversion rate and a $28.68 CPA in a study of approximately 20,000 accounts, while Average Ad Strength RSAs showed a 12.65% conversion rate and a $12.43 CPA.^[3] Adding weak, repetitive copy to boost Ad Strength from Good to Excellent can actively harm conversion performance. Add assets only when they contribute a new, meaningful message.

8. Mixing unrelated propositions in one RSA

An RSA that simultaneously promotes enterprise software, a free consumer trial, and a physical retail location will generate incoherent combinations. Keep each RSA internally thematic: one clear offer, one audience, one landing page.

9. Ignoring AI-generated assets after enabling text customization

Once text customization is active under AI Max, Google generates and refreshes assets at least every 48 hours.^[4] Failing to monitor the asset details report means inaccurate claims, expired promotions, or non-compliant language may serve without review. Schedule a monthly asset audit as a standing process.

10. Short, under-utilized descriptions

Descriptions of 0–50 characters show an 8.90% CTR and a $9.61 CPA in benchmark data, while descriptions of 61–70 characters show a 12.33% CTR, a 9.21% conversion rate, and a 307.58% ROAS.^[3] Use the available 90-character description space purposefully. A 40-character description that only restates the headline wastes one of the four available description slots.

13. What Changed Recently (Last 30 Days)

The following changes are current as of October 2026. No changes to core RSA character or asset limits have been confirmed in the last 30 days.^[4]

ACA renamed and migrated to AI Max as text customization

The most significant operational change is the transition of automatically created assets (ACA) into AI Max for Search under the label text customization. Migration began in September 2026 and is ongoing. Campaigns previously using ACA may now expose text customization settings and related AI Max controls at the campaign level.^[4] Advertisers should:

• Audit affected campaigns to confirm URL settings, brand language, and generated-asset behavior are correct under the new framework.
• Verify that any exclusions or restrictions previously applied to ACA have been correctly transferred to text customization settings.

New asset preview function

Google has introduced a preview function that displays up to 10 sample AI-generated headlines and descriptions before an advertiser opts into text customization. Previews generate in under 30 seconds and reflect any configured text guidelines.^[13] This feature materially reduces the risk of enabling text customization without knowing what copy will serve.

Text guidelines: term exclusions and messaging restrictions

Google is introducing and testing new text guideline controls that allow advertisers to exclude specific terms and restrict messaging concepts in AI-generated copy. The Ads API now supports programmatic term_exclusions and messaging_restrictions parameters.^[4] For accounts in regulated industries—financial services, healthcare, legal, insurance—these controls should be configured before enabling text customization.

Ad Strength now incorporates generated assets

When text customization is enabled, Ad Strength evaluates advertiser-created and Google-generated headlines and descriptions together.^[4] This means an RSA that previously rated Good may rate Excellent after enabling text customization—not because the human-authored copy improved, but because generated assets filled coverage gaps. Interpret post-migration Ad Strength changes with this context in mind.

Generated asset refresh cadence confirmed at 48 hours

Google has confirmed that AI-generated assets are reviewed and refreshed as needed at least every 48 hours, including when landing-page content changes.^[4] For advertisers running time-sensitive promotions, this means a landing-page change—such as extending a sale deadline—will typically be reflected in generated copy within 2 business days. Do not rely on this cadence as a guarantee; manually monitor the asset details report during promotional periods.

No change to RSA format limits

As of October 2026, the core RSA limits remain unchanged: up to 15 headlines at a maximum of 30 characters each, and up to 4 descriptions at a maximum of 90 characters each. Google continues to display up to 3 headlines and up to 2 descriptions per impression.^[5]

Limited Ad Serving update

Google’s updated Limited Ad Serving guidance, which affects new or infrequently served advertisers, now explicitly states that advertisers should use specific language rather than generic ad copy and landing-page content.^[4] This reinforces the existing best practice of avoiding vague benefit claims and generic CTAs, and it has direct implications for accounts that rely heavily on placeholder or templated copy.

---

References

1. [1] https://support.google.com/google-ads/answer/7684791/about-responsive-search-ads?hl=en support.google.com
2. [2] https://support.google.com/google-ads/answer/6167122?hl=en support.google.com
3. [3] https://www.optmyzr.com/blog/google-rsa-performance-study/ www.optmyzr.com
4. [4] https://scalevise.com/resources/google-ads-ai-max-migration-broad-match-aca-dsa/ scalevise.com
5. [5] https://support.google.com/google-ads/answer/7684791?hl=en support.google.com
6. [6] https://spideraf.com/articles/improve-google-ads-ctr spideraf.com
7. [7] https://developers.google.com/google-ads/api/docs/responsive-search-ads/overview developers.google.com
8. [8] https://support.google.com/google-ads/answer/9921843?hl=en support.google.com
9. [9] https://digitelia.co/case-studies/google-ads-ecommerce-5-7x-roas/ digitelia.co
10. [10] https://services.google.com/fh/files/misc/responsive_search_ads_a_guide_to_writing_ads_tha… services.google.com
11. [11] https://www.optmyzr.com/blog/google-ad-strength-study/ www.optmyzr.com
12. [12] https://support.google.com/google-ads/answer/7331111?hl=en support.google.com
13. [13] https://www.seroundtable.com/google-ads-text-customization-asset-previews-42264.html?utm_s… www.seroundtable.com
14. [14] https://support.google.com/google-ads/answer/9438230?hl=en support.google.com
15. [15] https://support.google.com/google-ads/answer/7507883?hl=en-AU&visit_id=639261592179992792-… support.google.com