Responsive Search Ads are the only Search ad format in Google Ads. You provide up to 15 headlines (30 characters each) and 4 descriptions (90 characters each), and Google’s AI assembles combinations in real time for each searcher. Your job isn’t to write one perfect ad — it’s to give Google a diverse, high-quality set of assets that work independently and in any combination.
This changes how ad copy works at a fundamental level. You’re not writing a single message. You’re writing a system of modular components that Google tests across thousands of combinations to find what converts. The advertisers who treat RSAs like “expanded text ads with extra slots” — writing 15 variations of the same headline — are the ones underperforming.
This guide covers how to structure your 15 headlines across distinct messaging angles, description writing that converts, pinning strategy, asset-level performance optimization, and the testing methodology that turns average ad copy into consistently high-performing campaigns.
The 30/90 Character Reality
Every copy decision happens within hard constraints: 30 characters per headline (including spaces), 90 characters per description (including spaces). Google silently truncates anything over the limit. On mobile, Google may show only one or two headlines, which means every single headline must function as a complete, compelling message on its own.
Write every headline as if it could appear alone in Position 1. Don’t write headlines that only make sense as a sequence (“The Best Way” / “To Save Money” / “Starts Here”). Google assembles combinations randomly — sequential headlines create nonsensical ads when separated.
The 15-Headline Framework: Eight Angles That Cover Everything
Most RSA underperformance comes from writing 15 variations of the same message. If your headlines all say versions of “Expert PPC Management” / “Professional PPC Services” / “Top PPC Agency,” Google is choosing between three weak variations instead of testing genuinely different messages.
Fill your 15 headline slots across these eight distinct angles:
Keyword Headlines (3 headlines)
Include your primary and secondary keywords. These establish relevance and improve Quality Score. Match the searcher’s language directly.
Examples: “Custom Window Installation” / “Residential Window Replacement” / “New Windows for Your Home”
Benefit Headlines (3 headlines)
State what the customer gets — not what you do, but what changes for them.
Examples: “Cut Energy Bills by 30%” / “Warmer Winters, Cooler Summers” / “Lower Heating Costs Year-Round”
Feature/Proof Headlines (2 headlines)
Specific, quantifiable claims about your product or service. Numbers outperform vague claims because they’re more credible and faster to process.
Examples: “Rated 4.9★ by 2,000+ Customers” / “25-Year Warranty Included”
Social Proof Headlines (2 headlines)
Customer counts, review ratings, awards, certifications. Third-party validation builds trust faster than self-promotion.
Examples: “Trusted by 10,000+ Homeowners” / “BBB A+ Rated Since 2015”
Urgency/Offer Headlines (2 headlines)
Time-sensitive offers, seasonal promotions, limited availability. Google’s countdown customizer can dynamically show remaining time.
Examples: “Free Installation — Ends Friday” / “Spring Sale: 20% Off All Windows”
CTA Headlines (1 headline)
Direct action language. Tell the searcher exactly what to do next.
Examples: “Get Your Free Quote Today” / “Schedule a Free Consultation”
Differentiator Headlines (1 headline)
What makes you different from competitors. The one thing you do that others don’t.
Examples: “Same-Day Installation Available” / “No Subcontractors — Our Own Crews”
Brand Headlines (1 headline)
Your company name or branded phrase. Builds recognition for searchers who see your ads repeatedly.
Examples: “WindowPro — Since 1998” / “Acme Windows & Doors”
This framework ensures Google has genuinely diverse messaging angles to test rather than minor rewrites of the same idea. Each angle serves a different persuasion function, and the AI learns which combinations convert best for each type of searcher.
Writing Descriptions That Convert
You get 4 descriptions at 90 characters each. Descriptions expand on your headlines and provide the supporting detail that turns interest into clicks.
Description 1 — your core value proposition. Explain what you offer and why it matters. Include your primary keyword naturally. “Professional window installation with lifetime warranty. Free in-home estimates — no obligation.”
Description 2 — address objections or add proof. Tackle the most common reason someone would hesitate. “Licensed & insured. Financing available. 4.9★ Google rating from 2,000+ verified reviews.”
Description 3 — highlight a specific offer or differentiator. Give a concrete reason to click now. “Spring promotion: 20% off all installations booked this month. Free energy audit included.”
Description 4 — alternative angle for different audiences. If your ad group targets multiple intent types, give Google a description that serves each. “Compare options with our online pricing calculator. No salespeople, no pressure.”
Each description must stand alone and make sense paired with any headline combination. Avoid descriptions that repeat what your headlines already say — the description’s job is to add information, not echo it.
Pinning Strategy: Control Without Killing Flexibility
Pinning locks a headline or description to a specific position. Used carefully, it ensures your most important message always appears. Used excessively, it recreates the rigid ETAs that RSAs were designed to replace.
Recommended approach:
Pin 2-3 of your strongest headlines to Position 1. This guarantees your most compelling message leads, while giving Google flexibility to test variations within that position.
Leave Positions 2 and 3 unpinned. Let Google’s AI test which combinations drive the highest click-through and conversion rates.
Don’t pin descriptions unless legally required (e.g., a mandatory disclaimer that must appear).
What happens when you over-pin: if you pin one headline to each of the three positions, you’ve essentially recreated a single static ad with no testing flexibility. Google can only show that one combination, which eliminates the learning advantage RSAs are designed to provide.
The exception: if regulatory requirements or brand guidelines mandate specific language always appear, pinning is appropriate. But even then, pin 2-3 variants to the same position rather than a single headline, so Google retains some testing room.
Ad Strength vs. Ad Rank: An Important Distinction
Ad Strength is Google’s rating (Incomplete → Poor → Average → Good → Excellent) of your RSA’s asset diversity and quantity. It evaluates whether you’ve provided enough varied headlines, included keywords, and covered different messaging angles.
Ad Strength does not affect Ad Rank. Google has confirmed this explicitly. A “Poor” rated RSA can outperform an “Excellent” rated RSA if the copy is more relevant and converts better. Ad Strength measures input breadth — how much variety you gave Google. It does not measure whether those inputs are good.
Use Ad Strength as a checklist during RSA creation (did you provide enough diverse assets?), but don’t chase “Excellent” ratings by adding filler headlines that dilute your messaging quality. Six strong, distinct headlines outperform fifteen mediocre ones, even if the latter scores higher on Ad Strength.
Asset-Level Performance: The Data That Changed Everything
Since mid-2025, Google provides click and conversion data per individual headline and description within an RSA. You can now see which specific headlines drive clicks, which drive conversions, and which are never shown.
Google rates each asset as “Best,” “Good,” “Low,” or “Learning.” The metrics that matter are conversion rate and cost per conversion per asset — not CTR in isolation. A headline with high CTR but low conversion rate is attracting clicks that don’t convert, which wastes budget.
Optimization workflow:
Check asset performance monthly after accumulating sufficient data (at least 1,000 impressions per asset).
Replace “Low” rated assets with new variations that test a different angle — not a rewrite of the same message.
Keep “Best” rated assets active. Don’t replace winners.
Assets stuck in “Learning” after 30+ days likely aren’t getting enough impressions. Consolidate ad groups or reduce the number of assets to give each one more testing opportunity.
Run at least 2 RSAs per ad group. This gives Google more assets to test while allowing you to compare performance between RSAs.
Automatically Created Assets (ACAs): When to Trust AI
Google now generates AI headlines and descriptions automatically based on your landing page content, existing ad copy, and account data. ACAs are enabled by default in many accounts.
When ACAs help: they fill gaps in your asset pool, surface relevant messaging you hadn’t considered, and test a broader range of combinations than manual assets alone.
When to override ACAs: when they produce generic headlines that don’t differentiate you, when they include claims you can’t substantiate, or when they conflict with your brand voice. Review ACAs regularly in your asset report — don’t assume AI-generated headlines are performing well just because they’re running.
You can disable ACAs at the campaign level if you need complete control over ad messaging. But for most accounts, leaving them enabled while monitoring performance is the better approach — they expand your testing pool without additional creative effort.
Ad Extensions (Assets) That Multiply Performance
Extensions add information below your ad at no extra cost and increase your ad’s visual footprint on the results page.
Sitelinks — additional links to specific pages (pricing, features, contact, case studies). Use 4-8 sitelinks per campaign. Each sitelink should point to a genuinely different, useful page.
Callouts — short phrases (25 characters max) highlighting key selling points. “Free Shipping” / “No Contract Required” / “24/7 Support” / “Money-Back Guarantee.” Add 4-6 callouts.
Structured Snippets — predefined headers (Types, Services, Brands, etc.) with a list of values. “Services: Installation, Repair, Replacement, Maintenance.” Add at least 4 values per header.
Call extensions — display your phone number. Particularly effective for service businesses where phone calls are the primary conversion action.
Location extensions — show your business address. Essential for businesses with physical locations targeting local searchers.
Price extensions — display product or service pricing directly in the ad. Useful for pre-qualifying clicks — users who see the price and still click are more likely to convert.
Extensions improve CTR by giving users more reasons and pathways to engage. They also increase your ad’s size on the results page, pushing competitors further down.
Testing Methodology
What to Test
Test one variable at a time within the RSA framework:
Headline angles — does a benefit headline outperform a social proof headline in Position 1?
Offer framing — does “20% Off” outperform “Save $50” for the same discount?
CTA language — does “Get a Free Quote” outperform “Request Pricing”?
Description approach — does an objection-handling description outperform a feature-focused one?
How to Test
Use Google Ads Experiments: duplicate your campaign, change only the RSA copy (keep keywords, bids, and audiences identical), run a 50/50 traffic split, and let it run until Google signals statistical significance. Minimum recommended volume: 10,000 impressions per month.
For faster informal testing within a single RSA, monitor asset-level performance ratings. Replace underperforming assets and measure the impact on overall RSA metrics.
Testing Cadence
Review asset performance monthly. Replace “Low” assets quarterly. Refresh all creative every 3-6 months to prevent fatigue. Even high-performing RSAs eventually decline as competitors adjust and audience behaviors shift.
Common Mistakes
Writing 15 headlines that say the same thing. “Expert PPC Management” / “Professional PPC Services” / “Top PPC Agency” gives Google three weak variations, not three distinct messages. Each headline should contribute unique information.
Ignoring character limits. Headlines over 30 characters get truncated silently. Descriptions over 90 characters get cut off. Count characters (including spaces) before submitting. AI-generated copy frequently exceeds limits — always verify.
Misaligning ad and landing page. If your headline promises “20% Off All Windows,” the landing page must show the 20% discount immediately. Mismatch kills conversion rates and damages Quality Score.
Chasing Ad Strength instead of performance. Adding filler headlines to reach “Excellent” dilutes your messaging quality. Six strong headlines outperform fifteen weak ones.
Never checking asset-level data. Weak headlines can run for months without being noticed if you only look at campaign-level metrics. Check individual asset performance monthly.
Ignoring mobile. Google may show only one headline on mobile. If your strongest message is in Position 3, mobile users may never see it. Pin your best headline(s) to Position 1.
Frequently Asked Questions
How many headlines should I write for an RSA?
Provide 8-15 headlines covering at least 5 distinct messaging angles. You don’t need to fill all 15 slots with filler — 10 strong, diverse headlines outperform 15 weak, repetitive ones. The goal is giving Google genuinely different messages to test, not hitting a number.
Should I use AI to write my Google Ads copy?
AI tools are excellent for generating first drafts, brainstorming headline variations, and producing options you might not have considered. But AI-generated copy must be reviewed for character limits (AI frequently exceeds 30/90 character constraints), factual accuracy, brand voice consistency, and differentiation from generic claims. Use AI to accelerate production, then apply human judgment to ensure quality.
Does Ad Strength affect my ad performance?
Not directly. Google has confirmed Ad Strength does not influence Ad Rank. It’s a diagnostic tool that evaluates whether you’ve provided enough diverse assets, not whether those assets are effective. Focus on conversion performance data rather than Ad Strength ratings.
When should I pin headlines in an RSA?
Pin 2-3 strong headlines to Position 1 to ensure your best message always leads. Leave Positions 2 and 3 unpinned for testing flexibility. Over-pinning (one headline per position) eliminates RSA’s core advantage. Only pin descriptions if legally required.
How often should I refresh my ad copy?
Review asset-level performance monthly. Replace “Low” performing assets as data accumulates. Refresh all creative every 3-6 months. Even winning copy eventually fatigues as competitive dynamics and audience behaviors shift. Treat ad copy as a continuous optimization loop, not a one-time setup.
What’s the single most important thing in Google Ads copy?
Relevance to the searcher’s intent. A headline that directly matches what the user is looking for — their problem, their desired outcome, their specific search query — outperforms clever wordplay, emotional triggers, and promotional offers. Match the intent first, then layer in persuasion through benefits, proof, and urgency.
