Responsive search ads are the only search ad you can still build. How Google assembles 15 headlines and 4 descriptions, what Ad Strength does and does not measure, and when pinning helps more than it hurts.
Last updated: November 15, 2026
In This Guide
- Executive Summary
- How RSAs Work
- Headlines & Descriptions
- Ad Strength
- Pinning
- Writing RSAs That Perform
- RSAs & AI Max
- Account Structure
- Common Mistakes
- What Changed Recently
- References
Executive Summary
Responsive Search Ads (RSAs) are the only search ad format available to advertisers in Google Ads standard Search campaigns as of November 2026. Expanded Text Ads were sunset on June 30, 2022, and cannot be created or edited, though legacy ETAs may still serve in some accounts.[1] Every active Search campaign requires at least one RSA, and best practice calls for at least two per ad group, each at Good or Excellent Ad Strength and each pointing to a unique final URL.[2][3]
Seven principles govern high-performance RSA management in 2026:
- Asset breadth matters more than asset count. Fill as many of the 15 headline and 4 description slots as possible, but only with genuinely distinct messages. Redundant or near-duplicate assets consume slots without adding testing value.
- Ad Strength is a creative diagnostic, not an auction signal. It measures asset quantity, diversity, and keyword relevance. It has no effect on Ad Rank or Quality Score. Good is the practical target; chasing Excellent by padding weak headlines wastes creative effort.
- Pin sparingly and strategically. Pinning restricts the combinations the system can test, reducing learning and typically lowering Ad Strength. Reserve pinning for legal disclaimers, regulated claims, and mandatory brand language. When you must pin, provide two to three unique assets per pinned position so the system retains rotation flexibility.
- Write for the user first, the algorithm second. Each headline and description should stand alone as a coherent, credible message. Headlines may now serve in sitelink-adjacent placements, so orphaned fragments perform poorly.
- AI Max for Search extends RSA reach but changes the rules. Its three components—Search Term Matching, Text Customization, and Final URL Expansion—can override pinned assets, select alternative landing pages, and generate novel copy. Test it with controls, not by default.
- The September 2026 automatic upgrades are live. Campaigns using legacy campaign-level broad match or standalone Automatically Created Assets have been or are being migrated to AI Max. Audit campaign settings now to verify your configurations were not silently changed.
- Measure business outcomes, not platform scores. Evaluate RSA performance using conversion rate, cost per conversion, conversion value, and lead quality. Use Ad Strength and asset-level reporting as diagnostic inputs, not as success metrics.
How Responsive Search Ads Work
An RSA is a machine-learning-assembled ad unit. The advertiser provides a pool of approved headlines and descriptions; Google’s systems assemble and serve combinations from that pool, learning over time which combinations are most likely to generate a qualified click for a given query and user context.[2]
The assembly process at each impression involves several factors:
- Query signals: The search term, its match type, and inferred intent.
- User context: Device, location, time of day, audience signals, and prior behavior where available.
- Ad space: The number of headline and description slots available in the specific auction placement.
- Expected performance: Google’s real-time prediction of which combination is most likely to achieve the campaign’s objective.
- Pinning constraints: Any advertiser-defined restrictions that require specific assets to appear in specific positions.
This is not random rotation. Google describes RSAs as testing different combinations and learning which perform best.[2] Combinations that perform well receive more delivery over time; combinations that underperform are shown less frequently. The practical implication is that the quality and variety of your asset pool directly determines the ceiling of what the system can optimize toward.
RSAs replaced Expanded Text Ads as Google’s standard format because they expose far more combinatorial surface area to testing than any fixed ad layout. A fully loaded RSA with 15 headlines and 4 descriptions creates tens of thousands of possible headline-and-description combinations, far more than any manual A/B test could evaluate at meaningful statistical confidence for most accounts.
One structural nuance matters for campaign architects: RSAs operate at the ad level within an ad group. Each RSA has its own final URL, its own asset pool, and its own performance history. This means two RSAs in the same ad group can test meaningfully different value propositions, messaging angles, or landing pages without requiring separate ad groups or campaigns.
Headlines and Descriptions: Limits and How Google Assembles Them
Asset Limits
| Asset type | Maximum per RSA | Minimum required | Character limit per asset | Slots shown per impression |
|---|---|---|---|---|
| Headlines | 15 | 3 | 30 characters | Up to 3 |
| Descriptions | 4 | 2 | 90 characters | Up to 2 |
| Display path fields | 2 | 0 | 15 characters each | Included in display URL |
These limits are fixed as of November 2026.[2][3] Google also allows up to three campaign-level headline assets and two campaign-level description assets, but those campaign-level assets are not currently included in Ad Strength calculations and do not substitute for a complete RSA asset pool.[4]
How Google Assembles a Served Combination
At each impression, Google selects up to 3 headlines and up to 2 descriptions from the available pool. The exact number shown varies by device, available screen space, and other serving conditions; not every impression shows the maximum 3 headlines or both descriptions. Headlines can appear in any order unless pinned, and descriptions can appear in either slot unless pinned.[2][3]
Google applies several rules during assembly:
- Redundant or highly similar assets are unlikely to appear together in the same served combination.
- Assets are evaluated for relevance to the specific query, so a keyword-rich headline may be favored for closely matched searches.
- Not every asset will appear in every impression; some assets in a 15-headline pool may serve rarely if the system consistently finds better combinations for the account’s dominant queries.
Display Path Fields
RSAs include two optional path fields appended to the display URL, each up to 15 characters. The display URL is derived from your final URL’s domain; path fields let you add descriptive text such as /Services/Plumbing to signal the destination without changing the actual URL. Path fields do not need to match the literal URL path of the final URL and are not guaranteed to appear in every impression.[4]
Use path fields to reinforce the offer or category. A financial services advertiser might use Retirement and Planning for a rollover IRA ad group. Keep path text specific and readable; avoid keyword stuffing or generic terms like Click or Here.
Campaign-Level Text Assets: What They Are and What They Are Not
Google now allows up to 3 campaign-level headlines and 2 campaign-level descriptions to be associated with RSAs across a campaign.[4] These assets can serve alongside RSA assets but carry a critical limitation: they are not included in Ad Strength calculations for the individual RSA.[4] Do not use campaign-level assets as a shortcut to improving Ad Strength. Build each RSA with a complete, diverse asset set at the ad level first.
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
Ad Strength: What It Measures and What It Does Not
What Ad Strength Measures
Ad Strength is Google’s creative diagnostic rating for RSAs. It evaluates three dimensions:[2][3]
- Quantity: How many headline and description slots are populated.
- Diversity: How distinct the messages, lengths, and themes of those assets are from one another.
- Keyword relevance: Whether some assets reflect the keywords and intent signals associated with the ad group.
The five rating levels are:
| Rating | Typical cause | Recommended action |
|---|---|---|
| Incomplete | Fewer than the required minimum assets | Add at least 3 headlines and 2 descriptions immediately |
| Poor | Few assets, high duplication, or very low keyword relevance | Add distinct assets; include relevant keyword language |
| Average | Moderate asset count with some repetition | Diversify messages; add benefit, offer, and CTA headlines |
| Good | Strong asset count with meaningful variety | Practical target for most RSAs; validate with performance data |
| Excellent | Full or near-full asset pool with high diversity and relevance | Aspirational; achievable without padding if messages are genuinely distinct |
What Ad Strength Does Not Measure
Ad Strength is explicitly not an input to Ad Rank, Quality Score, or the auction.[2] An RSA with a Poor rating does not receive a lower Ad Rank penalty; an RSA with Excellent does not receive a higher ranking bonus. Google has confirmed this directly.[2]
Ad Strength also does not measure:
- Conversion rate or conversion quality
- Landing page relevance or experience (that is captured separately in Quality Score)
- Message accuracy, persuasive strength, or brand compliance
- The actual click-through or conversion performance of an RSA
The 15% Statistic: Interpret It Carefully
Google reports that advertisers who improve Ad Strength from Poor to Excellent see on average 15% more clicks and conversions.[2] This is a correlation from Google’s internal data, not a causal guarantee, and it reflects averages across a large and diverse advertiser population. The improvement likely reflects the fact that ads with more diverse, relevant assets generally give the system more to work with, not that the Ad Strength label itself drives performance. Do not interpret it as a commitment that any individual account will see 15% more conversions by reaching Excellent.
The Right Target: Good, Not Excellent at Any Cost
Treat Good as the practical target for most RSAs. Reaching Excellent is worthwhile when you can supply genuinely distinct additional assets, but padding the last few headline slots with weak or repetitive copy to reach Excellent actively harms performance: it dilutes the asset pool with low-quality combinations and wastes slots that could carry meaningful messages.[3]
Asset-level performance reporting in Google Ads (ratings of Best, Good, Low, Learning, or Unrated for individual assets) is a more actionable signal than the aggregate Ad Strength score. Replace Low-rated assets with new angles rather than simply adding more assets.
Pinning: When to Use It and When to Avoid It
How Pinning Works
Pinning assigns a headline or description to a specific position so it appears whenever that position serves. Available pinning positions are:[2][5]
| Position | Always shows? | Recommended for mandatory messages? |
|---|---|---|
| Headline 1 | Yes, when the ad serves | Yes — strongest guarantee |
| Headline 2 | Yes, when the ad serves | Yes — strong guarantee |
| Headline 3 | No — optional slot | No — not guaranteed to appear |
| Description 1 | Yes, when two descriptions serve | Yes — use for essential body copy |
| Description 2 | No — optional slot | No — not guaranteed to appear |
If multiple assets are pinned to the same position, Google rotates among those pinned assets in that slot.[2][5] This is the preferred approach when pinning is necessary: it preserves some rotation and testing ability rather than permanently locking in a single asset.
Why Pinning Reduces Performance and Ad Strength
RSA optimization depends on combinatorial testing. Each pinned position removes that slot from the system’s mixing logic. As more positions are pinned, the number of valid combinations falls sharply. The effects include:[2][3]
- Fewer combinations available for Google to test against different queries
- Reduced ability to personalize the served combination by user context
- Lower Ad Strength, particularly when pinned assets are similar or duplicated
- Strong unpinned assets are blocked from appearing in pinned positions regardless of their expected performance
Google’s June 2026 guidance explicitly advises against pinning identical or very similar text to the same slot.[3] Pinning three near-identical brand variants—”Shop Acme,” “Buy Acme,” “Get Acme”—to Headline 1 provides almost no testing value and reduces Ad Strength without delivering material brand control.
An Important AI Max Interaction
Pinning behavior changes materially when AI Max for Search is enabled. If Final URL Expansion is active, pinned RSA assets may not be respected when Google selects an alternative landing page. If both Text Customization and Final URL Expansion are enabled together, pinning may be overridden.[1][5] If a mandatory message must always appear without exception, do not enable Final URL Expansion for that campaign, and verify that Text Customization settings do not substitute generated copy for pinned assets.
When to Pin
Pin only when a specific message must appear in a controlled position to satisfy a legal, regulatory, or mandatory business requirement. Legitimate use cases include:[2]
- Legal disclaimers or required disclosures (for example, FINRA-required language for financial advertisers)
- Regulated claims that must appear prominently under FTC or FDA guidelines
- Brand names that must appear in Headline 1 per internal governance policy
- Required product or service wording mandated by a licensing authority
- Essential offer conditions whose omission would make the ad misleading
When Not to Pin
- Preference for a particular phrasing that is not legally or contractually required
- Desire to control the “feel” of the ad without a specific compliance need
- To ensure a promotional message appears in every impression when that promotion has no legal requirement
- To compensate for a small or undiversified asset pool
The Recommended Hybrid Approach
For most advertisers, the optimal pinning configuration is:[2][3]
- Pin 2–3 unique, genuinely distinct assets to Headline 1 for mandatory brand or compliance messages.
- Leave Headlines 2 and 3 unpinned to maximize combinatorial testing.
- Pin Description 1 only if regulatory language must appear in the body copy.
- Leave Description 2 fully unpinned.
This approach ensures the mandatory message always appears while preserving meaningful rotation and optimization across all other positions.
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 InstagramWriting RSAs That Perform
Start With a Complete Asset Strategy Before Writing a Single Line
Before writing any copy, map out the full 15 headline slots by message category. This prevents duplication and ensures Google has genuinely distinct content to assemble. A recommended framework:
| Headline slots | Message category | Example (home security advertiser) |
|---|---|---|
| 1–3 | Primary keyword and service category | Home Security Systems, Monitored Alarm Systems, Wireless Security Cameras |
| 4–6 | Key benefits and outcomes | 24/7 Professional Monitoring, Real-Time Alerts on Your Phone, Deter Burglars Instantly |
| 7–9 | Features, speed, or convenience | DIY Install in 30 Minutes, No Long-Term Contracts, Works With Alexa and Google |
| 10–12 | Offer, price, or call to action | Equipment From $99, Get a Free Quote Today, Start for $1 This Month |
| 13–15 | Differentiators, proof, or guarantees | A+ BBB Rating Since 2005, 500,000+ Homes Protected, 60-Day Money-Back Guarantee |
| Description slots | Message category | Example |
|---|---|---|
| 1 | Value proposition and primary benefit | Protect your family with professional-grade security. 24/7 monitoring, mobile alerts, and fast police dispatch—starting at $99. |
| 2 | Major features or offer detail | No contracts. No hidden fees. Includes HD cameras, door and window sensors, and a free professional installation consult. |
| 3 | Objection handling or proof | Rated #1 for customer satisfaction by PCMag three years running. Cancel anytime with no early termination fee. |
| 4 | Strong call to action and next step | Get your free customized security quote in 2 minutes. No obligation. Speak with a security advisor today. |
Seven Writing Rules With Practical Application
1. Make every asset independently legible. Any headline may appear alongside any two other headlines and either description in any order. Write each asset so it communicates a coherent message even without knowing what will appear next to it. This is especially important now that RSA headlines can serve in sitelink-adjacent placements.[6] A fragment like “Award-Winning” is too thin; “Award-Winning Customer Support” works as a standalone claim.
2. Lead with the user’s specific goal. Match the intent implied by the ad group’s keyword theme in at least 2–3 headlines. If your ad group targets “workers compensation insurance for small business,” at least two headlines should reflect that exact need—not a generic umbrella like “Business Insurance Solutions.” Specificity improves relevance scores and user response.
3. State a concrete, verifiable differentiator. Vague claims like “Best Service” or “Great Prices” carry no weight. Specific claims do: “Free 2-Day Delivery Over $49,” “Licensed in All 50 States,” “Rated 4.9/5 by 12,000 Customers.” If you cannot substantiate a claim on the landing page, do not write it into the RSA.
4. Use specific calls to action. “Get a Free Quote,” “Schedule a Consultation,” “Compare Plans,” and “Start a Free Trial” tell the user exactly what happens next and attract more qualified clicks than generic directives. Match the CTA to the funnel stage: an informational query warrants “Learn More” or “Compare Options”; a transactional query warrants “Buy Now” or “Add to Cart.”
5. Reflect the landing page precisely. The price, offer, terminology, and action stated in the ad must appear on the landing page. Mismatches increase bounce rate, damage Quality Score, and in regulated categories can trigger policy violations. If your RSA says “Free Consultation,” the landing page must offer a free consultation, not a paid assessment with a buried free-call option.
6. Vary character length across headlines. Mix short headlines (10–18 characters) with medium (19–24 characters) and near-maximum (25–30 characters) lengths. This gives Google flexibility to assemble clean layouts across devices where available character space varies.
7. Test themes, not near-duplicates. A second RSA should explore a meaningfully different angle: a different value proposition, a different user segment, a different offer, or a different landing page. Two RSAs with the same 15 headlines re-ordered provide no incremental testing value. Consider divergent hypotheses: one RSA emphasizes price and speed; the second emphasizes expertise and trust.
Keywords in RSA Copy: The Right Balance
Include your ad group’s primary keyword or close variant in 2–3 headlines—enough to establish relevance and support quality score signals, but not so many that every headline becomes a keyword insertion. Keyword-heavy assets often read as mechanically as Dynamic Keyword Insertion and perform poorly when the system assembles them together. Descriptions are where you typically elaborate on benefits rather than repeat keyword phrases.
Headlines Must Now Survive Sitelink Placement
As of 2026, Google’s enhanced asset flexibility allows up to two RSA headlines to serve in space previously reserved for sitelinks.[6] This is a structural change to how RSA assets are reused across the ad unit. Audit every headline in your RSA to verify it works as a standalone informational fragment without the supporting context of adjacent headlines and descriptions. Headlines like “Call Us Today” or “And Much More” fail this test; “Free Overnight Shipping Over $75” passes it.
RSAs and AI Max for Search
What AI Max for Search Is
AI Max for Search is a campaign-level feature suite introduced by Google that extends how Search campaigns match queries, generate text variants, and select landing pages around your RSA assets. It is not a new ad format; your RSA remains the foundation. AI Max adds three capabilities on top of it:[1]
| AI Max component | What it does | Dependency |
|---|---|---|
| Search Term Matching (STM) | Expands query coverage beyond your keyword list to additional relevant searches using AI signals | None — can be enabled independently |
| Text Customization | Formerly Automatically Created Assets; generates additional RSA headlines and descriptions tailored to individual queries using your landing pages, keywords, and existing assets as source material | None — can be enabled independently |
| Final URL Expansion (FUE) | Replaces the RSA’s final URL with another page on your domain when that page is expected to better match the query; may generate copy to match the new landing page | Requires Text Customization to be active |
The September 2026 Automatic Upgrades
Beginning September 1, 2026, Google began automatically migrating Search campaigns using legacy campaign-level broad match or standalone Automatically Created Assets to AI Max.[3] The migration defaults differ by legacy configuration:
- Campaigns using campaign-level broad match: Migrated to AI Max with Text Customization and Final URL Expansion opted out by default; Search Term Matching is enabled.[3]
- Campaigns using standalone Automatically Created Assets: Search Term Matching and Text Customization are enabled by default; Final URL Expansion is opted out by default.[3]
Google also blocked creation of new legacy campaign-level broad match and standalone ACA configurations from August 3, 2026.[3] If you have not audited your campaigns since September 2026, do so immediately. Verify your campaign settings, check which AI Max components are enabled, review URL exclusions, and confirm that asset sources and brand term controls are correct after any automatic migration.
Text Customization: Constrained Expansion, Not Unrestricted Copywriting
Text Customization draws inputs from your domain’s crawlable pages, your existing ads, your keywords, and your manually supplied RSA assets.[1] Google’s AI then generates query-specific headline and description variants from those inputs. It is constrained expansion because the quality, accuracy, and compliance of the source material directly determines the quality, accuracy, and compliance of what gets generated. Outdated landing-page copy, unsupported promotional claims, or unapproved product language on your site can become inputs to generated assets.
Practical controls for Text Customization:
- Keep all landing pages current, accurate, and crawlable. Remove expired promotions, outdated pricing, and discontinued products.
- Review generated assets in Google Ads reporting before scaling spend. Check both the asset text and the source context Google used to generate it.
- Use ad-level and campaign-level exclusions for sensitive messaging categories.
- Separate brand campaigns, regulated campaigns, and promotional campaigns from prospecting campaigns where broader automation is acceptable.
Final URL Expansion and Pinning: A Critical Conflict
When FUE is active, Google can select a landing page different from the RSA’s final URL and generate corresponding copy for that page. This directly conflicts with pinning: pinned assets from the original RSA may not be respected when FUE selects a different URL.[1][5] If a mandatory message—a legal disclaimer, a regulated disclosure, a required brand term—must appear in every impression without exception, Final URL Expansion must be disabled for that campaign. This is not a workaround; it is Google’s own stated guidance.[1]
Enhanced Asset Flexibility: RSA Headlines in Sitelink Space
Google’s enhanced asset flexibility change allows up to two RSA headlines to serve in space previously reserved for sitelinks.[6] When AI Max’s Text Customization is also active, this creates a broader pool of eligible headlines—including generated ones—that may appear in that placement. For brand-sensitive accounts, test this in an isolated campaign first and verify that every headline serving in sitelink space is compliant, coherent, and consistent with your brand standards.
Reported Performance Impact
Google reports that advertisers already using Search Term Matching saw on average a 7% increase in conversions or conversion value at a similar CPA or ROAS after enabling Text Customization and Final URL Expansion together.[1] This figure is from Google’s internal global data for 2026 and excludes Retail advertisers. It is an average, not a forecast for any individual account. Treat it as a test hypothesis: results will vary by conversion volume, landing-page quality, industry, brand constraints, attribution model, and the amount of incremental query coverage available.
Decision Framework: When to Enable and When to Hold Back
Enable the full AI Max suite when:
- Conversion tracking is reliable and conversion volume is sufficient to support automated bidding and AI learning.
- Landing pages are well structured, current, and consistent across all pages Google may select.
- Incremental query reach is commercially desirable and the account can tolerate some variation in search terms.
- CPA, ROAS, or conversion-value bidding strategies can absorb additional variance in the short term.
- No legal or contractual obligation requires a specific URL or specific ad text to appear in every impression.
Use a controlled or partial rollout when:
- The campaign has regulated, brand-sensitive, or legally controlled messaging requirements.
- Landing pages contain mixed-quality content, user-generated content, or non-approved promotional language.
- URL selection must remain fixed for analytics, affiliate tracking, contractual, or compliance reasons.
- Search term expansion could expose the brand to irrelevant or reputationally damaging query categories.
A practical rollout sequence: enable AI Max on one representative non-brand campaign, retain a comparable control campaign where possible, and evaluate performance using conversion quality, marginal CPA or ROAS, search term relevance, landing page engagement metrics, and any brand safety incidents. Do not evaluate solely on aggregate conversion count. Outside the US, feature availability, data-use regulations (particularly GDPR in the EU), and content generation policies may differ; verify local Google Ads documentation before enabling FUE or Text Customization in regulated markets.
“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 worksAccount Structure: How Many RSAs and How to Organize Them
The Minimum Standard
Best practice is at least two RSAs per ad group, each at Good or Excellent Ad Strength and each with a unique final URL.[2][3] Two RSAs in the same ad group allow Google to rotate between meaningfully different creative strategies while giving asset-level performance reporting enough data to distinguish which approaches resonate with your audience.
Ad Group Thematic Tightness
RSA performance depends heavily on the alignment between keywords in an ad group and the assets in the RSA. Write RSAs for the specific intent of that ad group’s keyword theme, not as a generic message that attempts to cover the entire campaign. An ad group targeting “emergency plumber near me” should have RSA headlines that reflect urgency, availability, and fast dispatch—not the same general headlines as an ad group targeting “bathroom remodeling services.”
Tight thematic ad groups also make Ad Strength easier to achieve legitimately: when the keyword set is narrow and coherent, the relevant headlines and descriptions write themselves more naturally and without forced repetition.
RSA Count: Quality Over Quantity
A common mistake is creating three, four, or five RSAs in a single ad group as a substitute for investing in strong individual RSA asset pools. Multiple near-duplicate RSAs fragment the data needed for learning and make asset-level reporting harder to interpret. The recommended hierarchy is:
- Primary RSA: Strongest keyword relevance, most complete asset pool, points to the most targeted landing page. Target Good or Excellent Ad Strength.
- Test RSA: Meaningfully different value proposition, messaging angle, or landing page. Use this to test a competing hypothesis—price vs. trust, features vs. outcomes, free trial vs. money-back guarantee.
- Third RSA (optional): Only justified when the first two have sufficient data to show a performance pattern and you have a genuinely distinct third angle or landing page to test.
Organizing Campaigns With AI Max in Mind
The September 2026 AI Max rollout makes campaign-level organization more consequential. Recommended structural separation:
- Brand campaigns: Keep tightly controlled, with pinning for brand name in Headline 1, Final URL Expansion disabled, Text Customization reviewed carefully or disabled, and negative keyword lists protecting branded terms from non-brand campaigns.
- Regulated or legally sensitive campaigns: Disable FUE, disable or restrict Text Customization, audit generated assets regularly, and maintain explicit URL controls.
- Prospecting and non-brand campaigns: These are the most appropriate testing ground for the full AI Max suite where conversion tracking and landing-page quality support it.
Landing Page as a Structural Element
Each RSA’s final URL should lead to the most relevant page for that ad group’s intent. For accounts with well-structured websites, this usually means category or product pages rather than the homepage. When FUE is enabled, Google may also select subcategory or detail pages; ensure every crawlable page on your domain that Google might select meets quality, compliance, and brand standards.
Common Mistakes to Avoid
- Treating Ad Strength as a performance metric. Ad Strength does not predict conversion rate, CPA, or revenue. Advertisers who optimize for Excellent Ad Strength by adding weak, repetitive assets often see their actual business metrics worsen because the ad pool is diluted with low-quality combinations. Always cross-reference Ad Strength improvements with conversion data before declaring success.[2]
- Over-pinning for messaging preferences. Pinning more than one or two positions eliminates most of the combinatorial value that makes RSAs superior to fixed ad formats. If you find yourself pinning five or more headlines across all three positions, you have effectively rebuilt an Expanded Text Ad in an RSA shell. Remove preference-based pins and rely on asset-level performance reporting to surface your strongest messages organically.
- Pinning a single asset to position 3 or description 2 for critical messages. Positions Headline 3 and Description 2 are not guaranteed to appear in every impression. Any message that must always be present belongs in Headline 1, Headline 2, or Description 1.[2]
- Writing near-duplicate headlines to fill slots. Variants like “Affordable Home Insurance,” “Cheap Home Insurance,” and “Low-Cost Home Insurance” occupy three headline slots while contributing only one idea. The system is unlikely to serve these together anyway, and the slot that could carry a unique benefit or differentiator is wasted. Each headline should represent a genuinely different message category.
- Assuming campaign-level text assets improve Ad Strength. They do not. As of November 2026, campaign-level headlines and descriptions are explicitly excluded from Ad Strength calculations.[4] Build complete RSAs at the ad level.
- Enabling Final URL Expansion on campaigns with mandatory pinned messages. FUE can override pinned assets when it selects an alternative landing page. If a pinned asset carries a legal or regulatory requirement, FUE must be disabled for that campaign or the pinned message is not guaranteed to appear.[1]
- Not auditing campaigns after the September 2026 AI Max migration. Automatic migration may have enabled or changed settings that alter how your RSA assets, URLs, and query matching behave. Any account that has not been reviewed since the migration window should be audited before spending continues at scale.
- Writing RSA headlines that only make sense in context. With enhanced asset flexibility allowing RSA headlines to serve in sitelink-adjacent placements, context-dependent fragments fail in isolation. Every headline should communicate a coherent, standalone claim.[6]
- Running one RSA per ad group. A single RSA provides no competitive creative testing and limits the data available for asset-level learning. The minimum standard is two RSAs per ad group, each with a unique final URL and a meaningfully different creative strategy.[2]
- Measuring RSA success by click-through rate alone. CTR is an upstream signal. Final success metrics are conversion rate, cost per conversion, conversion value, qualified lead rate, and revenue. An RSA with a lower CTR that generates higher-intent clicks and more conversions outperforms one with a higher CTR and lower conversion quality in every business-relevant dimension.
What Changed Recently (Last 30 Days)
The following developments reflect the most material changes to RSA management documented in available official sources as of November 2026. This section will require ongoing review as Google continues to evolve the platform.
Campaign-Level RSA Text Assets Now Available
Google now supports up to 3 campaign-level headline assets and 2 campaign-level description assets that can associate with RSAs across a campaign.[4] These provide incremental asset coverage without requiring individual RSA edits, but they carry an important limitation: they are explicitly excluded from Ad Strength calculations for the individual RSA.[4] Account managers should not use these as a shortcut to improving Ad Strength ratings or as a substitute for complete ad-level asset pools.
Ad Strength Formally Excludes Campaign-Level Text Assets
Google has clarified in its current support documentation that campaign-level headlines and descriptions are not currently factored into Ad Strength.[4] This is a meaningful operational update: it closes a potential misunderstanding that adding campaign-level assets would improve an RSA’s diagnostic rating. Each RSA must still be evaluated and built on its own merits at the ad level.
AI Max Pinning Override Behavior Formally Documented
Google’s support documentation now explicitly states that pinned RSA assets may not be respected when Final URL Expansion is active, and that pinning is also not respected when URL inclusions are provided.[1][5] This is not a new technical behavior, but its formal documentation in support resources is new and consequential for accounts managing regulated or legally sensitive campaigns under AI Max settings. Any campaign with legally mandatory pinned messaging should have Final URL Expansion disabled.
RSA Headlines Can Now Serve in Sitelink Placements
Google has announced and is rolling out enhanced asset flexibility that allows existing RSA headlines to appear in space previously reserved for sitelinks.[6] This change means RSA headlines can receive additional serving opportunities outside the traditional 3-headline ad layout. The practical implication is that every RSA headline must now be written to function as a coherent standalone message, because it may be surfaced without the supporting context of adjacent headlines or descriptions. Advertisers should audit existing RSA headline pools and rewrite any context-dependent fragments.
AI Max September 2026 Migration Is Live
The automatic upgrade of legacy campaign-level broad match and standalone Automatically Created Assets campaigns to AI Max began on September 1, 2026.[3] New creation of those legacy configurations was blocked from August 3, 2026.[3] The migration window means some accounts will have had settings automatically changed without explicit advertiser action. Priority action items for any account not reviewed since September:
- Open each Search campaign and verify which AI Max components are enabled or disabled.
- Check whether Final URL Expansion was opted in during migration (it should be opted out by default for both migration paths, but verify).[3]
- Review search term reports for any new query patterns introduced by Search Term Matching.
- Confirm that asset sources for Text Customization are limited to approved, current landing pages.
- Validate that brand term exclusions and negative keyword lists survived the migration intact.
No New RSA Format Changes or ETA Reinstatement
Available documentation for the past 30 days shows no newly deprecated RSA features, no reinstatement of Expanded Text Ads, and no changes to the core RSA asset limits of 15 headlines and 4 descriptions.[2] The platform trajectory remains toward broader AI-controlled assembly, greater campaign-level asset reuse, reduced reliance on rigid pinning, and more opportunities for existing RSA assets to serve across multiple ad components.
Related reading
- Building a Google Ads Strategy
- Google Ads Optimization: Which Lever To Pull
- Google Ads Keyword Research: Finding Keywords That Convert
- Google Ads Lead Generation: Lead Form Assets
Or skip the work and get a Search campaign set up by a Google Ads specialist.
References
- [1] https://support.google.com/google-ads/answer/11777352?hl=en support.google.com
- [2] https://support.google.com/google-ads/answer/7684791/about-responsive-search-ads?hl=en support.google.com
- [3] https://support.google.com/google-ads/answer/9438230?hl=en-0 support.google.com
- [4] http://support.google.com/google-ads/answer/7046795 support.google.com
- [5] https://support.google.com/google-ads/answer/7684791?pZ5YM= support.google.com
- [6] https://support.google.com/google-ads/answer/9497781?hl=en-tt support.google.com

