Responsive Search Ads Best Practice Guide

Last updated: September 15, 2026

Executive Summary

Responsive Search Ads (RSAs) are the only search ad format advertisers can build in Google Ads as of September 2026. Expanded text ads were retired on June 30, 2022, and while legacy ETAs remain visible in accounts, they can only be paused, resumed, or removed—not edited or newly created.^[1] Every live search ad is now an RSA, which means mastering the format is no longer optional.

The following principles govern high-performing RSAs in the current environment:

• Asset volume matters, but diversity matters more. Supply up to 15 headlines and 4 descriptions, but ensure each asset conveys a meaningfully different message. Padding an ad with near-duplicate headlines to chase a higher Ad Strength rating is counterproductive.^[2]
• Ad Strength is a diagnostic, not an auction signal. It measures asset quantity, diversity, and keyword relevance. It does not affect Ad Rank or Quality Score. The practical target is Good; reaching Excellent by adding weak or repetitive assets is not worth the tradeoff.^[3]
• Pin sparingly and strategically. Pinning restricts combination testing and typically lowers Ad Strength. Pin only assets that legally or contractually must appear in a specific position—brand names, regulated claims, legal disclaimers—and pin 2 to 3 unique assets per pinned slot so the system retains some optimization room.^[4]
• Maintain at least two RSAs per ad group at Good or Excellent Ad Strength, each with a unique final URL. This gives Google’s machine learning enough variation to optimize across queries while keeping copy relevant to the landing page.^[5]
• AI Max for Search is not optional for much longer. From September 1, 2026, Google began automatically upgrading Search campaigns using campaign-level broad match or standalone automatically created assets to AI Max. Treat it as a controlled expansion layer on top of your RSAs rather than an all-or-nothing automation switch.^[6][7]

How Responsive Search Ads Work

An RSA is best understood as an asset pool combined with a machine-learning assembly engine. You provide the creative inputs—headlines, descriptions, display path fields, and a final URL—and Google’s system assembles and tests combinations at auction time based on the query, user signals, device, location, and competitive context.^[2][8]

At each impression, Google selects up to 3 headlines and up to 2 descriptions from your pool. The selection is not random: Google’s models favor combinations predicted to achieve the best outcome for that specific query and user context. Over time, higher-performing combinations appear more frequently, and weaker ones are deprioritized. This is the core value proposition of the format—continuous multivariate testing without requiring manual A/B test setup.^[8]

The ad’s visible URL is composed of your domain plus two optional path fields, each limited to 15 characters. These path fields do not change the destination; they only customize what the user sees in the ad. The final URL determines where the click actually lands.^[2][8]

Expanded text ads served a fixed combination of text every time. RSAs invert that model: the copy is fluid across impressions, the landing page is fixed per RSA, and the system learns which fluid combinations drive results. That shift has significant implications for how copywriters should think about the asset set—individual headlines are no longer independent ads; they are ingredients in hundreds of potential combinations.

One important behavioral nuance: Headline 3 and Description 2 are not guaranteed to appear in every impression. Ad serving constraints, screen size, and format eligibility mean some impressions show only 2 headlines or 1 description. Any message that must always appear should be in Headline 1, Headline 2, or Description 1—and even then, use pinning only if the requirement is genuine.^[8]

Headlines and Descriptions: Limits and How Google Assembles Them

The table below summarizes the hard limits for RSA assets and display paths as of September 2026.

Asset Type	Minimum Required	Maximum Allowed	Character Limit Per Asset	Shown Per Impression
Headlines	3	15	30 characters	Up to 3
Descriptions	2	4	90 characters	Up to 2
Display Path 1	0 (optional)	1	15 characters	When set, always shown
Display Path 2	0 (optional)	1	15 characters	When set, always shown

The 30-character headline limit is strict and includes spaces. Google will reject headlines that exceed it. In practice, write to around 25 to 28 characters to avoid truncation on smaller screen sizes, while keeping enough characters to convey the message clearly.

The 90-character description limit is also hard. Descriptions appear below the headlines and carry more explanatory weight, so use all available space purposefully—state a proof point, expand on an offer, or deliver a persuasive call to action.

How Google Assembles Combinations

Google’s assembly logic aims to maximize predicted performance for each auction. The system learns which headline and description combinations perform best for particular query patterns and user contexts, and it surfaces those combinations more often over time. Because headlines can appear in any order across position 1, 2, and 3, each individual headline must make sense in any position and alongside any other headline in your pool. A headline that only makes sense when read after another specific headline will produce incoherent ads in many combinations.^[2][8]

There is one important asset flexibility development from 2026 to be aware of: Google has introduced an enhanced asset flexibility change that allows up to two RSA headlines to serve in space previously reserved for sitelinks. This raises the stakes for message hierarchy—if a promotional or benefit-oriented headline surfaces in a sitelink-adjacent position, it needs to stand alone and represent the brand appropriately.^[8] Review your full headline pool with that context in mind.

Practical Implications for Copywriters

• Write every headline as a standalone unit. It may appear first, second, or third, or in a sitelink-adjacent space.
• Vary the purpose of each headline across the pool: keyword-focused, benefit-focused, offer-focused, brand-focused, call-to-action-focused, and social-proof-focused headlines all serve different query intents.
• Descriptions pair with other descriptions, so write each one as a complete persuasive unit rather than a continuation of the previous description.
• Use both path fields whenever possible. They contribute to perceived relevance and can reinforce the keyword or category without adding to character counts in the main copy.

Ad Strength: What It Measures and What It Does Not

Ad Strength is a five-level diagnostic rating assigned to each RSA: Incomplete, Poor, Average, Good, and Excellent. Google calculates it based on three factors: the quantity of assets provided, the diversity of messaging across those assets, and the relevance of assets to the keywords in the ad group.^[3]

Ad Strength is a setup quality signal. It is not an auction input. It does not influence Ad Rank, Quality Score, or how often the ad enters auctions. An RSA rated Poor competes in the same auction with the same Ad Rank mechanics as an RSA rated Excellent. The rating tells you whether the creative foundation is strong enough for the system to do effective optimization work—it does not tell you whether the ad is winning impressions or driving conversions.^[3]

Google has reported that advertisers who improve Ad Strength from Poor to Excellent see an average of 15% more clicks and conversions.^[3] This is a correlation from Google’s own aggregate data, not a controlled experiment. The underlying driver of that lift is almost certainly better creative diversity—more unique, relevant assets—rather than the rating itself. Chase the underlying quality, not the score.

What Each Level Practically Means

Ad Strength Level	Typical Cause	Recommended Action
Incomplete	Fewer than the minimum assets (3 headlines / 2 descriptions)	Add remaining required assets immediately; the ad may not be eligible to serve
Poor	Minimum assets only, low diversity, repetitive wording	Add unique headlines and descriptions covering different message angles
Average	Adequate quantity but limited diversity or keyword relevance	Replace the weakest, most similar assets with genuinely differentiated copy
Good	Strong diversity and keyword relevance across most assets	This is the practical target; test asset swaps to maintain and incrementally improve
Excellent	High quantity, high diversity, strong keyword relevance across all assets	Appropriate target if achievable with genuinely strong copy; do not pad to reach it

What Ad Strength Does Not Measure

Ad Strength does not evaluate copy quality in a human editorial sense. An ad rated Excellent can still have weak value propositions, vague calls to action, or mismatched landing pages. It also does not evaluate conversion rate, click-through rate, or landing page relevance beyond keyword matching. Use Ad Strength alongside your actual performance metrics—CTR, conversion rate, cost per conversion, ROAS—and treat it as a signal of setup adequacy, not creative excellence.^[3]

The Pinning Effect on Ad Strength

Pinning assets to specific positions reduces the number of combination variants the system can test. Fewer testable combinations typically lower Ad Strength because the system interprets the constrained setup as lower optimization flexibility. If your Ad Strength is lower than expected and your account requires pinning for compliance reasons, that is an expected tradeoff, not a problem to solve by adding more pinned assets.^[4]

Pinning: When to Use It and When to Avoid It

Pinning assigns a specific headline or description to a fixed position—Headline 1, Headline 2, Headline 3, Description 1, or Description 2—so Google will only ever show that asset in that slot. It is the mechanism for guaranteeing message placement when business, legal, or compliance requirements demand it.^[2][4]

The core tradeoff is control versus optimization. Every pinned slot is a slot where Google cannot test alternatives. The more slots you pin, the fewer combination variants the system can assemble, and the lower your Ad Strength and testing efficiency will be.^[4]

The Updated June 2026 Pinning Guidance

Google’s June 2026 guidance made the pinning recommendation more specific in two ways. First, it explicitly advises against pinning identical or very similar text to the same slot—doing so provides the appearance of multiple pinned options while delivering no real variation for optimization. Second, it recommends pinning 2 to 3 genuinely unique assets to each pinned position when pinning is necessary, so the system can rotate among meaningful variants within the constraint.^[4]

For example, if your brand name must appear in Headline 1, pin two or three unique Headline 1 options that all include the brand name but differ in structure or accompanying message. This preserves the brand requirement while restoring some optimization room.

The Hybrid Pinning Approach

The recommended approach as of September 2026 is a hybrid strategy:

• Pin essential assets to Headline 1 (and Headline 2 or Description 1 if truly required) using 2 to 3 unique variants per pinned position.
• Leave Headline 2, Headline 3, and Description 2 unpinned wherever business and compliance rules allow.
• Never pin to force a preferred creative order when there is no genuine compliance or brand requirement. Aesthetic preference is not a valid pinning reason.

Appropriate vs. Inappropriate Pinning Use Cases

Scenario	Pin?	Notes
Legal disclaimer required in the ad copy	Yes	Pin to Headline 1 or Description 1; provide 2–3 compliant variants if wording allows
Brand name must appear in every impression	Yes	Pin to Headline 1 with 2–3 unique brand-name variants; leave other positions open
Regulated health or financial claim with exact wording	Yes	Pin exact required language; document the compliance rationale internally
Strong preference for a specific headline order	No	Creative preference does not justify restricting optimization; trust the system
Wanting the “best” headline always in position 1	No	The system already surfaces better-performing assets more frequently; pinning overrides that
Testing a new offer or message	No	Add as an unpinned asset and let performance data determine prominence

Auditing Your Pinning Footprint

Review the Position Pinning column in the ad-level asset report regularly. This column shows which assets are pinned, to which positions, and how frequently those assets are serving. If you see a pinned asset delivering a high share of impressions with below-average performance relative to unpinned assets, that is evidence the pin is actively suppressing better-performing combinations.^[9]

Writing RSAs That Perform

Effective RSA copywriting requires a different mental model from traditional ad writing. You are not writing one ad; you are building a structured creative system where every asset must function independently and in combination. The goal is to cover the widest possible range of search intent, user context, and message angle while maintaining coherence in every combination the system might assemble.

Headline Strategy

With 15 available headline slots and only 3 shown per impression, treat the pool as a deliberate message architecture. Assign each headline a role and ensure no two headlines fill the same role with near-identical wording. A working framework for the 15-headline pool:

• Keyword-focused (2–3 headlines): Include the primary keyword or close variants. These drive relevance signals and are especially important when Dynamic Keyword Insertion is not in use. Do not use the same keyword phrase in all three; vary the construction.
• Primary benefit (2 headlines): State what the user gets or achieves. Focus on outcomes, not features.
• Feature or differentiator (2 headlines): Highlight what makes the product, service, or offer distinct. Be specific—vague differentiators like “High Quality” add little value.
• Offer or incentive (2 headlines): Discounts, free trials, free shipping, guarantees. If the offer changes seasonally, these are the headlines to update.
• Social proof (1–2 headlines): Customer counts, ratings, years in business, certifications, or named awards. Specificity matters—”Over 12,000 Customers Served” outperforms “Trusted by Thousands.”
• Call to action (1–2 headlines): Direct, verb-led prompts. “Get a Free Quote Today,” “Shop the Sale Now,” “Start Your Free Trial.” Keep these short enough to work at 30 characters.
• Brand (1 headline): Your brand name alone, or brand plus a short tagline. Useful for reinforcing awareness, especially in competitive auctions.
• Problem or pain point (1 headline): Acknowledge the user’s challenge before stating the solution. This angle often performs well for high-intent queries.

Description Strategy

With only 4 description slots available and 2 shown per impression, descriptions carry more persuasive weight per character than headlines. Each description should be a standalone unit that adds value independent of which other description appears alongside it. Suggested allocation:

• Primary value proposition: The single most compelling reason to choose your product or service. Write this as if it is the only description the user will see.
• Proof point or credibility: Supporting evidence—data, certifications, guarantees, media mentions, customer results.
• Offer with urgency: Specific promotion details with any relevant deadline or constraint. If no offer exists, use a strong benefit expansion instead.
• Call to action with instruction: Tell users exactly what to do next and what they will get when they do it. “Call now for a free 15-minute consultation” is more actionable than “Contact us today.”

Asset-Level Rules for US Advertisers

• Never use all-caps except for recognized acronyms (e.g., “HVAC,” “IRS,” “APR”). Google’s editorial policies flag excessive capitalization.^[2]
• Do not include prices or promotional claims in descriptions that are inconsistent with the landing page. FTC guidelines apply to digital advertising, and price mismatches can generate both policy flags and consumer complaints.
• Use punctuation carefully. Exclamation marks are not permitted in headlines under Google’s current editorial policies.
• If you serve regulated categories—financial products, healthcare, legal services, housing, employment—review Google’s ad policies for that vertical before writing claims. Regulated category restrictions differ by state in some cases.^[2]

A Practical RSA Copy Checklist

• Are all 15 headline slots filled with unique, meaningfully different copy?
• Are all 4 description slots filled?
• Do at least 2–3 headlines include the primary keyword or a close variant?
• Does every headline make sense in position 1, position 2, and position 3 independently?
• Does every description work as a standalone unit?
• Are both path fields used to reinforce the category or keyword?
• Is the final URL the most relevant landing page for this ad group’s intent?
• Is the Ad Strength at Good or Excellent without relying on near-duplicate assets?
• Are pins limited only to genuine compliance or brand requirements?

RSAs and AI Max for Search

AI Max for Search is not a new campaign type. It is an optimization layer that sits inside existing Search campaigns and enhances how those campaigns match queries, generate or adapt ad text, and route traffic to landing pages. As of September 2026, it comprises three primary components: search term matching, text customization (the successor to automatically created assets), and final URL expansion.^[6][7]

RSAs remain the foundational ad format within AI Max campaigns. The RSA is still what gets assembled and served. AI Max changes the conditions under which that RSA is shown and allows the system to supplement or adapt the RSA’s text and destination based on query and user context.

The Three AI Max Components and How They Interact With RSAs

AI Max Feature	What It Does	Interaction With RSAs	Key Risk to Manage
Search Term Matching	Expands eligibility beyond your keyword list to capture additional relevant queries	RSA is shown on queries your keywords would not have matched; system selects best asset combinations for each query	Irrelevant or brand-unsafe query expansion; monitor search term reports closely
Text Customization	Uses landing page content and generative AI to create or adapt headlines and descriptions for specific queries	AI-generated text supplements or replaces RSA assets in some impressions; original RSA assets still anchor the format	Off-brand, factually incorrect, or policy-violating generated text; review asset reports for generated copy
Final URL Expansion	Routes clicks to a more contextually relevant page on your site rather than the RSA’s designated final URL	The landing page for some impressions may differ from the RSA’s final URL	Traffic sent to pages with different offers, pricing, or compliance disclaimers than the ad implied

The September 2026 Automatic Upgrade

Google began automatically upgrading Search campaigns to AI Max on September 1, 2026. Two populations of campaigns are affected:

• Campaigns using standalone automatically created assets: Upgraded to AI Max with both search term matching and text customization active by default.^[6][7]
• Campaigns using campaign-level broad match: Upgraded to AI Max with search term matching active by default; text customization is not automatically enabled for this group.^[7]

If your account falls into either category and you have not proactively reviewed your AI Max settings, you may already be running with expanded automation. Check campaign-level settings in the Google Ads UI and review the AI Max controls panel to confirm which features are active.

Reported Performance and How to Interpret It

Google reports an average 7% uplift in conversions or conversion value from enabling the full AI Max feature suite.^[10] This figure represents an average across a large and varied advertiser population. Actual lift will vary substantially based on landing page quality, conversion tracking completeness, query mix, creative asset quality, and how much of the original RSA asset pool is relevant to the expanded query set. Treat 7% as a directional benchmark, not a guaranteed outcome for your account.

How to Approach AI Max Practically

The most defensible approach for most US advertisers in late 2026 is incremental enablement with active monitoring:

• Start with clean RSA structure. AI Max layers on top of your RSAs, so weak or sparse creative inputs produce weaker AI-generated outputs. A well-structured RSA pool with 15 diverse headlines and 4 descriptions gives the system better source material.
• Enable search term matching first. It carries the lowest creative risk because your RSA assets are still what gets served—only the query eligibility expands. Monitor the search terms report weekly for irrelevant or brand-inappropriate queries and add negative keywords accordingly.
• Enable text customization only when your landing pages are reliable source material. The system pulls from landing page content to generate or adapt ad text. Outdated, inconsistent, or thin landing pages will produce low-quality generated assets. If your landing pages are strong and current, text customization can be valuable.
• Be cautious with final URL expansion if exact offer-to-landing-page alignment matters. Compliance-sensitive advertisers—those in financial services, healthcare, legal services, or any category with state-specific regulations—should evaluate final URL expansion carefully. Sending users to a page that does not reflect the ad’s stated terms is both a conversion rate problem and a potential regulatory issue.
• Use brand controls. AI Max includes brand safety and URL controls that let you restrict which domains or pages are eligible for final URL expansion and set boundaries on query matching. Use them.
• Run an incremental test before full rollout. Set up a control campaign alongside an AI Max-enabled variant, give both sufficient budget and time to accumulate statistical significance, and evaluate actual conversion volume and value—not just uplift percentages—before expanding automation account-wide.^[10]

The Enhanced Asset Flexibility Implication

The enhanced asset flexibility change that allows RSA headlines to serve in sitelink-adjacent positions adds a new review obligation for AI Max users. If text customization generates headlines that surface in those positions, you need to verify that those generated headlines are brand-appropriate, factually accurate, and policy-compliant when read in isolation. Build a regular asset review step into your account management cadence—at minimum monthly, and weekly during the initial AI Max ramp period.^[8]

Account Structure: How Many RSAs and How to Organize Them

The minimum viable RSA setup per ad group is two RSAs, each at Good or Excellent Ad Strength, each with a unique final URL.^[5] This is Google’s current stated guidance and reflects the logic that the system needs multiple creative variants to identify which combinations perform best across different queries and user contexts.

Why Two RSAs With Different Final URLs

Two RSAs per ad group serve distinct functions. The first RSA typically anchors the core offer or primary landing page for the ad group. The second RSA tests a different message angle, value proposition, or audience frame—and because it carries a unique final URL, it can route to a landing page optimized for that angle. Over time, performance data from both RSAs informs which combinations, messages, and landing pages drive the best results for the ad group’s query intent.

Using the same final URL for both RSAs wastes structural potential. If both ads send users to the same page, differentiation only happens at the ad copy level, not the full user journey. Differentiated final URLs allow true message-to-landing-page alignment testing.

Ad Group-Level Organization

RSAs perform best when the ad group’s keyword theme is tightly scoped. Broad, mixed-intent ad groups make it harder for the system to select relevant headline and description combinations, and they reduce the keyword relevance component of Ad Strength. Recommended structural principles:

• Keep ad groups to a single keyword theme or closely related cluster. If a keyword set spans meaningfully different user intents, split it into separate ad groups with dedicated RSAs.
• Write RSA headlines that reflect the specific language of the ad group’s keywords. Generic headlines that could appear in any ad group miss the relevance opportunity the format provides.
• Match each RSA’s final URL to the most relevant landing page for that ad group’s intent. A campaign selling both enterprise software and small-business software should not send all ad groups to the same homepage.
• Audit ad group RSA count and Ad Strength quarterly. Ad groups with only one RSA or with Average or Poor Ad Strength are leaving optimization capacity on the table.

When to Use More Than Two RSAs

Two RSAs is the recommended minimum, not a cap. In high-volume ad groups—those generating tens of thousands of impressions per month—a third RSA can provide additional creative diversity for the system to test. Use a third RSA to explore a substantially different message frame: a competitor comparison angle, a different audience segment (e.g., new customers vs. returning customers), or a seasonal variant.

However, adding a third RSA that closely mirrors the first two in messaging does not help the system. Additional RSAs only add value if they provide genuinely new creative angles and are supported by unique landing pages.

Account-Level Implications of AI Max on Structure

AI Max’s search term matching can blur the boundaries between ad groups by expanding query eligibility beyond your keyword list. If your account structure is already loose—broad ad groups with overlapping themes—AI Max amplifies that overlap. Clean up ad group structure before enabling AI Max at scale to avoid serving RSAs on queries that belong to a different, more relevant ad group.

Common Mistakes to Avoid

The following errors appear regularly in RSA accounts and are worth addressing explicitly, particularly for teams transitioning from expanded text ad workflows.

Treating Ad Strength as a KPI

Ad Strength is a setup diagnostic. Optimizing toward Excellent by adding headlines that repeat the same ideas with slightly different wording reduces the quality of the system’s combination testing without improving real business outcomes. Prioritize genuine creative diversity over the rating. If an ad is at Good with 12 meaningfully distinct headlines, adding 3 near-duplicate headlines to reach Excellent is net negative.^[3]

Over-Pinning

The most common structural error in RSA accounts is excessive pinning inherited from expanded text ad habits. Teams accustomed to controlling every word of every ad default to pinning all positions to preserve the familiar three-headline, two-description structure they used in ETAs. This eliminates the optimization benefit of the RSA format entirely. Unless compliance or brand guidelines require fixed placement, leave positions unpinned.^[4]

Duplicate or Near-Duplicate Headlines

Submitting 15 headlines that say essentially the same thing in slightly different words—”Get a Free Quote,” “Request a Free Quote,” “Free Quote Available”—signals low diversity to the Ad Strength system and provides no meaningful creative variation for the assembly engine to test. Write each headline as if it represents a distinct reason to click.^[2]

Ignoring the Asset Report

The ad-level asset report shows which individual headlines and descriptions are rated Best, Good, or Low based on observed performance relative to other assets in the same RSA. Advertisers who never review this report miss the primary signal for iterative copy improvement. Replace Low-rated assets with new approaches; preserve and expand on Best-rated asset themes.^[9]

Mismatched Landing Pages

An RSA that promises a specific offer—”50% Off This Week Only”—and lands the user on a generic homepage rather than an offer-specific page will underperform on conversion rate regardless of how well-constructed the RSA is. Landing page relevance affects Quality Score and actual conversion rate. Each RSA’s final URL should be the most specific, message-matched page available.^[2]

Running Only One RSA Per Ad Group

A single RSA gives Google’s system limited creative variance to test and provides no competitive comparison at the ad group level. Two RSAs with different final URLs and different message angles is the minimum structure for meaningful optimization.^[5]

Ignoring AI Max Activation After the September 2026 Upgrade

Campaigns automatically upgraded to AI Max will run with expanded automation active unless settings are reviewed and adjusted. Advertisers who are unaware of the upgrade may find their campaigns matching to broader query sets, generating text from landing pages, or routing traffic to unexpected URLs without realizing it. Check campaign settings now if you have not done so since September 1, 2026.^[6][7]

Assuming AI Max Always Outperforms Manual Settings

Google’s reported 7% average lift is an aggregate figure across a large population. For individual accounts—especially those in regulated industries, with narrow product catalogs, strict brand guidelines, or limited conversion tracking—AI Max’s full feature suite may not deliver positive incremental results. Test before scaling, and use brand controls and URL exclusions to manage risk during the test period.^[10]

What Changed Recently (Last 30 Days)

The most significant changes affecting RSA management in the period leading into September 2026 are the AI Max automatic upgrade rollout and the reinforced pinning guidance. Here is a concise summary of what changed, what stayed the same, and what to do now.

Area	What Changed	What Stayed the Same	Recommended Action
AI Max Upgrade (Sept 1, 2026)	Google began automatically upgrading Search campaigns using campaign-level broad match or standalone automatically created assets to AI Max[6][7]	RSAs remain the core ad format; manual campaign types without these features are not auto-upgraded	Audit campaign settings immediately; review AI Max feature toggles and enable brand controls before expanding reach
Text Customization	Now formally part of AI Max; terminology consolidated; auto-enabled for campaigns with automatically created assets[6]	The underlying behavior—generating or adapting ad text from landing pages—is unchanged; pinned RSA assets are still respected	Review generated asset reports; verify generated copy for accuracy, brand tone, and policy compliance
Pinning Guidance (June 2026)	Google explicitly advised against pinning identical or near-identical assets to the same slot; recommended 2–3 unique assets per pinned position[4]	Core pinning mechanics unchanged; pinned assets still appear in their assigned position; Headline 3 and Description 2 still not guaranteed every impression	Audit existing pins; replace single-asset pins with 2–3 unique variants; remove pins from positions without a genuine compliance or brand requirement
RSA Asset Limits	No change: 15 headlines / 4 descriptions maximum remains current[2]	All character limits and minimum requirements unchanged	No action required on limits; continue building to full 15/4 capacity with diverse assets
Enhanced Asset Flexibility	Up to two RSA headlines can now serve in space previously reserved for sitelinks[8]	Sitelinks remain a separate asset type; this change allows headline overflow into that space under certain conditions	Review all RSA headlines for standalone brand safety; ensure every headline reads appropriately in isolation
AI Max Planning and Measurement Tools	New tools rolling out in September 2026 to help advertisers model scale and evaluate performance effects of the AI Max migration[10]	Existing reporting, asset reports, and search term visibility remain available	Use the new planning tools to model expected query volume changes before enabling AI Max on high-spend campaigns

For account managers reviewing this article in September 2026, the immediate priority is confirming whether any managed campaigns were automatically upgraded to AI Max on September 1, and if so, reviewing the active AI Max feature set against each campaign’s brand, compliance, and performance requirements. The RSA format itself is stable—no changes to asset limits or auction mechanics—but the ecosystem around RSAs is evolving rapidly, and the automation defaults are more expansive than they were six months ago.

Outside the US, AI Max rollout timing and specific feature availability may differ by market. Google has indicated the September 2026 migration timeline applies primarily to US and major English-language markets; confirm current status for other regions via the Google Ads Help Center or your Google account representative before assuming equivalent behavior.

---

References

1. [1] https://searchengineland.com/google-to-sunset-expanded-text-ads-etas-in-june-2022-374153 searchengineland.com
2. [2] https://support.google.com/google-ads/answer/7684791?hl=en support.google.com
3. [3] https://support.google.com/google-ads/answer/9921843?hl=en support.google.com
4. [4] https://ppcnewsfeed.com/ppc-news/2026-06/google-ads-tips-pinning-headlines-descriptions/ ppcnewsfeed.com
5. [5] https://support.google.com/google-ads/answer/6167122?hl=en support.google.com
6. [6] https://blog.google/products/ads-commerce/dsa-upgrade-to-ai-max-2026/ blog.google
7. [7] https://ads-developers.googleblog.com/2026/08/migrate-campaign-level-broad-match-and.html ads-developers.googleblog.com
8. [8] https://developers.google.com/google-ads/api/docs/responsive-search-ads/overview developers.google.com
9. [9] https://support.google.com/google-ads/answer/9564897?hl=en support.google.com
10. [10] https://blog.google/products/ads-commerce/ai-max-new-features/ blog.google