Responsive Search Ads Guide: How to Write Better RSAs in 2026

Responsive Search Ads Guide: How to Write Better RSAs in 2026

What Are Responsive Search Ads

Responsive Search Ads (RSAs) are the standard ad format for Google Search campaigns. You provide multiple headlines and descriptions, and Google’s machine learning system assembles them into different combinations based on the search query, device, browsing history, and other signals. The system then optimises towards the combinations that perform best.

RSAs replaced Expanded Text Ads (ETAs) as the default creation format in June 2022. While existing ETAs continued to serve, no new ETAs can be created. This means every search advertiser — including those running campaigns for Singapore markets — needs to understand how to write effective RSAs to compete in Google Search.

The format allows up to 15 headlines (30 characters each) and 4 descriptions (90 characters each). Google selects up to 3 headlines and 2 descriptions to display in any given auction. The combinations are tested automatically, with the system learning over time which combinations generate the highest click-through rates, conversion rates, or other performance signals aligned to your bidding strategy.

For businesses looking for managed campaign support, our Google Responsive Search Ads services handle everything from ad copywriting to ongoing optimisation.

How RSAs Work

Understanding the mechanics of RSAs is essential for writing effective ads. The system is not random — it makes intelligent decisions based on available data.

Combination assembly. When a user searches a query that matches your keywords, Google selects a subset of your headlines and descriptions to show. The selection is influenced by the search query (relevant headlines are prioritised), the user’s device (shorter combinations may be preferred on mobile), and historical performance data (combinations that have previously generated clicks or conversions are favoured).

Learning period. When you create a new RSA, Google enters a learning period during which it tests different combinations to gather performance data. This period typically lasts a few weeks but varies based on impression volume. During this time, performance may fluctuate as the system experiments. Avoid making changes to the ad during this period, as it resets the learning process.

Auction-time signals. RSAs adapt in real time. Two users searching the same keyword may see different headline and description combinations based on their location, time of day, device, and previous browsing behaviour. This is the core advantage of RSAs over static ad formats — each impression is assembled to maximise relevance for that specific user.

Reporting limitations. Google reports on which headline and description combinations are shown most frequently and their relative performance, but the reporting is less granular than many advertisers would like. You can see which individual assets are rated “Best”, “Good”, or “Low”, but you cannot see the exact click-through rate for specific headline-description pairings. This opacity is a trade-off for the system’s automation.

For a broader understanding of how ad copy fits into campaign performance, see our Google Ads ad copy guide.

Writing Effective Headlines

Headlines are the most visible element of your ad. Each headline has a maximum of 30 characters, which is not much — every word must earn its place.

Include the focus keyword in at least 2 to 3 headlines. This ensures that when Google assembles a combination, at least one headline is likely to include the term the user searched for. Keyword-relevant headlines improve both click-through rate and Quality Score. For example, if you are targeting “responsive search ads,” include headlines like “Responsive Search Ads Agency” or “Expert RSA Management.”

Write headlines that work independently. Google may display your headlines in any combination and any order. Each headline must make sense on its own and when placed next to any other headline. Avoid headlines that are sequential (“Part 1 of Our Service…” and “Part 2 of Our Service…”) as they may not appear together.

Vary the message across headlines. Do not write 15 variations of the same point. Instead, cover different themes:

• Keywords and services: “Google Ads Management” / “PPC Campaign Services”
• 好处 “Increase Your ROAS” / “Lower Cost Per Lead”
• 可信度： “10 Years of Google Ads Experience” / “Google Partner Agency”
• Calls to action: “Get a Free Audit Today” / “Request a Proposal”
• Location: “Singapore-Based Agency” / “Serving SG Businesses”
• Differentiation: “No Lock-In Contracts” / “Transparent Monthly Reports”

This variety gives Google more options to assemble relevant combinations for different search intents.

Use all 15 headline slots. Google recommends using all available slots, and Ad Strength (more on this below) penalises ads that do not. More headlines mean more possible combinations, which gives the system more data to work with and more room to optimise.

Include numbers and specifics. Headlines with specific figures — “300+ Clients Served”, “ROI Up 47%”, “From S$500/Month” — tend to outperform vague alternatives. Numbers stand out in search results and communicate concrete value.

Test dynamic keyword insertion (DKI) in one or two headlines. Using the {KeyWord:Default Text} function inserts the user’s search term into the headline, improving relevance. However, do not rely on DKI alone — some search terms may produce awkward or grammatically incorrect headlines. Use DKI as a supplement, not a replacement for well-written static headlines.

Writing Strong Descriptions

Descriptions support the headlines with additional detail. Each description has a maximum of 90 characters, giving you more room to elaborate.

Expand on the headline promise. If a headline says “Lower Cost Per Lead”, the description should explain how: “Our data-driven approach to Google Ads optimisation reduces wasted spend and improves lead quality.” The description is where you provide the evidence behind the headline claim.

Include a clear call to action in at least one description. “Contact us today for a free campaign audit” or “Request a no-obligation proposal online.” Users need to know what to do next. While headlines often include short CTAs, descriptions allow for more specific instructions.

Use all 4 description slots. Like headlines, using all available slots gives the system more combinations to test. Cover different angles:

• Description 1: Core value proposition and service overview.
• Description 2: Social proof, credibility, or specific results.
• Description 3: Call to action with urgency or incentive.
• Description 4: Differentiation or secondary benefit.

Ensure descriptions work with any headline combination. This is the same principle as headlines: since Google assembles combinations dynamically, each description must complement any set of headlines without redundancy or contradiction.

Include relevant keywords naturally. Work the target keyword or close variants into descriptions where it reads naturally. Forced keyword stuffing degrades the user experience and can lower click-through rates even if it improves ad relevance scoring.

Pinning Strategy

Pinning lets you fix a specific headline or description to a specific position. A headline pinned to Position 1 will always appear as the first headline when the ad is shown. This gives you control over what appears — but it comes at a cost.

How pinning affects performance. Every pin reduces the number of combinations Google can test. If you pin headlines to all three positions and descriptions to both positions, you effectively create a static ad — eliminating the RSA’s core advantage. Google’s Ad Strength metric penalises heavy pinning for this reason.

When to pin. Pinning is justified in specific situations:

• Brand name visibility: Pin a headline containing your brand name to Position 1 if brand recognition is critical.
• Regulatory compliance: If your industry requires specific disclosures (financial services disclaimers, for example), pin those to ensure they always appear.
• Message coherence: If certain headlines only make sense in a specific position (a CTA that should always be last), pinning prevents nonsensical combinations.

Pin multiple assets to the same position. Instead of pinning one headline to Position 1, pin two or three. This tells Google “always show one of these in Position 1” while still allowing variation within that constraint. You maintain some control without fully eliminating the system’s ability to optimise.

Avoid pinning everything. The more you pin, the more the RSA behaves like a static ETA, and the less opportunity Google has to find high-performing combinations. As a rule, pin only when there is a clear business or compliance reason, and pin multiple assets per position when possible.

Understanding how pinning interacts with Quality Score is important. Our Google Ads Quality Score guide explains the factors that determine ad rank and cost per click.

Ad Strength and Optimisation

Ad Strength is Google’s rating of your RSA, ranging from “Poor” to “Excellent”. It is displayed during ad creation and in the Ads tab of your campaign. Google uses it to encourage best practices, but its relationship to actual performance is nuanced.

What Ad Strength measures. The rating is based on:

• Number of assets: Using all 15 headlines and 4 descriptions improves the score.
• Uniqueness: Headlines and descriptions that are distinct from each other score higher than repetitive ones.
• Keyword inclusion: Including keywords from the ad group in your headlines boosts the score.
• Pinning: Excessive pinning lowers the score.

Ad Strength is directional, not definitive. An ad with “Excellent” Ad Strength is not guaranteed to outperform one with “Good” Ad Strength. Google has stated that Ad Strength is a guide, not a performance predictor. Some advertisers have reported that ads with “Good” Ad Strength outperform “Excellent” ones because the actual ad copy resonated better with the audience, even if it did not tick every checkbox in Google’s rubric.

Aim for “Good” or “Excellent”. While Ad Strength is not a perfect predictor, ads rated “Poor” or “Average” are often genuinely underserving their potential. Use the recommendations to improve — add more headlines, diversify your messaging, include keywords — but do not sacrifice copy quality to chase the rating.

Review asset performance labels. In the Ads tab, Google assigns performance labels (“Best”, “Good”, “Low”, “Learning”) to individual headlines and descriptions. Replace “Low”-performing assets with new alternatives and monitor whether the replacements improve. This iterative process is more valuable than chasing Ad Strength alone.

Do not conflate Ad Strength with Quality Score. These are different metrics. Quality Score affects your ad rank and cost per click. Ad Strength is a creative guideline. Both matter, but they measure different things.

Testing and Iteration

RSAs require a different approach to testing than static ad formats. You cannot run a clean A/B test between two headline-description combinations because you do not control which combination is shown.

Test at the ad level, not the asset level. Create two RSAs in the same ad group with different messaging strategies. For example, one RSA focused on price-related messaging and another focused on quality-related messaging. After sufficient data (typically 1,000+ impressions per ad), compare click-through rates, conversion rates, and cost per conversion to determine which strategy performs better. For guidance on structuring tests, our A/B testing guide covers the methodology in detail.

Rotate underperforming assets. Every two to four weeks, review the asset performance labels. Replace any headline or description labelled “Low” with a new alternative. Do not replace more than two or three assets at once — changing too many resets the learning period and makes it difficult to attribute improvements.

Test one variable at a time when replacing assets. If you replace a headline, change only the messaging angle or the phrasing. Do not simultaneously change the CTA, the keyword, and the benefit. Isolating variables helps you learn what works.

Allow sufficient data before drawing conclusions. RSAs need time and impressions to stabilise. Do not judge an ad’s performance after three days with 200 impressions. Wait until each RSA has at least 1,000 impressions and 30+ conversions (if optimising for conversions) before making decisions. In low-volume campaigns, this may take several weeks.

Use experiments for structural tests. Google Ads Experiments let you split traffic between a control and a variant. Use this feature to test significant changes — different RSA structures, new landing pages, or alternative bidding strategies — with statistical rigour. Our Google Ads services include experiment design and analysis for clients who want to test systematically.

Document learnings. Maintain a simple log of what you tested, what you changed, and what happened. Over time, this log becomes a playbook of what works for your specific account, industry, and audience. Patterns emerge: certain benefit statements, CTA styles, or proof points consistently outperform others.

Common RSA Mistakes

Avoid these common errors that reduce RSA effectiveness.

Writing headlines that are too similar. If 10 of your 15 headlines are minor variations of the same message (“Best Google Ads Agency”, “Top Google Ads Firm”, “Leading Google Ads Company”), you are not giving the system much to work with. Each headline should contribute a different angle or piece of information.

Ignoring mobile formatting. On mobile devices, Google may show only 2 headlines instead of 3, and only 1 description instead of 2. If your key message is in the third headline or second description, mobile users may never see it. Ensure that your strongest messaging appears in assets likely to be shown in Position 1.

Writing descriptions that repeat headlines. If a headline says “Free Campaign Audit” and a description says “Get a free audit of your campaign today,” the user is reading the same message twice. Descriptions should expand on headlines, not echo them.

Over-pinning. Pinning every asset to a fixed position eliminates the RSA’s ability to optimise. Only pin when there is a clear reason (brand compliance, regulatory requirements), and pin multiple assets per position to maintain flexibility.

Neglecting to update ads. RSAs are not set-and-forget. Seasonal messaging, expired promotions, outdated pricing, and “Low”-performing assets should all be reviewed and updated regularly. A quarterly review cycle is the minimum; monthly is better for active campaigns.

Using the same RSA across all ad groups. Each ad group targets different keywords with different intent. The RSA in each group should reflect that intent. A generic ad used across multiple ad groups will underperform tailored ads written specifically for each group’s keyword theme.

Chasing Ad Strength at the expense of copy quality. Adding a 15th headline just to reach “Excellent” Ad Strength, when that headline adds no meaningful message, is counterproductive. Quality and relevance matter more than filling every slot with mediocre content. Aim for at least 10 to 12 strong headlines and all 4 descriptions as a practical minimum.

常见问题

How many RSAs should each ad group have?

Google recommends one RSA per ad group as a starting point. You can have up to three enabled RSAs per ad group, which is useful for testing different messaging strategies. However, having more than three active RSAs splits impression data too thinly, making it difficult for the system to learn effectively. For most Singapore advertisers, one to two RSAs per ad group is the practical sweet spot — one primary ad and one variant for testing.

Does Ad Strength directly affect ad rank or cost per click?

No. Ad Strength is an editorial guideline, not an auction signal. Your ad rank is determined by your bid, Quality Score (which includes expected click-through rate, ad relevance, and landing page experience), and the expected impact of ad extensions. A “Good” Ad Strength ad with excellent Quality Score will outperform an “Excellent” Ad Strength ad with poor Quality Score in the auction. That said, following Ad Strength recommendations often improves the signals that feed into Quality Score, so the two are indirectly related.

Can you still use Expanded Text Ads alongside RSAs?

Existing ETAs that were created before June 2022 can still serve alongside RSAs. You can pause and re-enable them, but you cannot create new ETAs or edit the text of existing ones. Over time, Google is shifting optimisation resources towards RSAs, and many advertisers have found that their RSAs outperform legacy ETAs. If your ETAs are still performing well, there is no need to pause them — let them run alongside your RSAs and let the data decide.

What is the minimum number of headlines and descriptions for an RSA?

Google requires a minimum of 3 headlines and 2 descriptions to create an RSA. However, using the minimum severely limits the system’s ability to optimise. Google recommends at least 8 to 10 unique headlines and all 4 descriptions. Providing 11 to 15 headlines gives the system the most material to work with, but only if each headline adds genuine value. Five excellent headlines and five mediocre ones will not outperform ten strong ones.

How long should you wait before optimising an RSA?

Allow at least two to three weeks for the learning period, assuming the ad group receives a reasonable volume of impressions (at least 1,000 per week). For low-volume campaigns (fewer than 500 impressions per week), the learning period can stretch to four to six weeks. Making changes during the learning period — swapping headlines, adjusting pins, changing descriptions — resets the learning process. Be patient, let the data accumulate, and then make informed changes based on asset performance labels and overall ad metrics.