How to Write Google Responsive Search Ads That Convert (With AI Prompts)

You loaded three headlines, two descriptions, hit save, watched Google flag the ad "Average," and told yourself you would fix it later. Later never came. That ad is now the reason your click rate is flat and your cost per click keeps creeping up: you handed the algorithm nothing to test.

To write responsive search ads that convert, treat the ad as a modular kit, not a single piece of copy: write all 15 headlines (30 characters each) and all 4 descriptions (90 characters each) so any headline can sit next to any other and still read cleanly. Cover distinct angles across those assets, keyword, benefit, offer, proof, urgency, and CTA, then pin only your one strongest keyword headline to position 1 and let Google combine the rest. Variety and standalone clarity are what earn a strong Ad Strength and a served ad.

What is a responsive search ad, really?

A responsive search ad (RSA) is a single ad unit where you supply a pool of interchangeable headlines and descriptions, and Google assembles them into different combinations in real time, testing which mix wins for each search. You are not writing one ad; you are writing the parts of many possible ads.

A responsive search ad is a text ad built from up to 15 headlines and 4 descriptions that Google mixes and matches automatically, showing up to 3 headlines and 2 descriptions per impression based on what it predicts will perform best for that query. Your job is not to write the perfect line. It is to hand the machine a deep, varied, non-repetitive supply of good lines it can recombine.

That distinction changes how you write every asset. In a classic ad you controlled the exact sequence. In an RSA, headline 7 might show first, or third, or not at all, and it has to make sense in every one of those spots. As of 2026 the RSA is the only standard search text format Google offers, so getting it right is the single biggest lever you have on click-through rate and conversions.

What are the specs and limits I have to work inside?

Before the framework, know the box you write inside. Spaces count.

You can technically launch with 3 headlines and 2 descriptions. Do not. Google recommends all 15 headlines and all 4 descriptions, and Ad Strength drops toward "Poor" with fewer. Every headline you skip is a combination Google cannot test.

Thirty characters is roughly four to six words, so there is no room for filler. Cut "your" and "our" almost everywhere: "Boost Sales Today" (17) says as much as "Boost Your Sales Today" (21). Treat every character as paid inventory, because it is.

The 15/4 RSA framework: how do I fill every slot without repeating myself?

The failure mode is obvious once you see it: people write three genuinely different headlines, then pad to fifteen with near-duplicates. Google sees the redundancy, has nothing new to test, and rates the ad weak. The fix is to assign each headline a job first, mapping your 15 across six angle buckets so no two lines do the same work.

1. Keyword-match headlines (aim for 3 to 4). These mirror the search term as closely as your product allows and make your ad feel like the answer to the query. If someone searches "project management software," a headline reading "Project Management Software" wins relevance and Quality Score.

2. Primary benefit headlines (3 to 4). The core outcome the buyer wants: results, not features. "Ship Projects 2x Faster" beats "Gantt Charts Included."

3. Offer and price headlines (2 to 3). The commercial hook: "Start Free, No Card Needed," "14-Day Free Trial," "Plans From $12/Month." These convert bottom-funnel searchers who are comparing.

4. Proof and trust headlines (1 to 2). Credibility that lowers risk. Only use numbers you can stand behind. A line like "Rated 4.8 by 2,000 Teams" is fine if it is literally true; otherwise write "Trusted by Growing Teams." Never invent a stat to fill a slot.

5. Urgency and scarcity headlines (1 to 2). A reason to act now, when honest: "Offer Ends Sunday," "Limited Spots This Month." If nothing is genuinely time-bound, skip this bucket rather than fake it.

6. CTA headlines (1 to 2). A direct instruction: "Get Your Free Demo," "Try It Free Today." These give Google a clean, action-oriented line to close a combination.

For the 4 descriptions, run the same logic at 90 characters, where you have room for a full thought. Give each a distinct job: (1) benefit plus offer, (2) proof plus a specific feature, (3) objection handling ("No contracts, cancel anytime"), (4) a CTA with a soft deadline. Four descriptions, zero overlap.

Two rules bind the whole thing together. First, every headline must stand alone and pair with any other, because Google controls the order. Imagine any two of your lines stacked; if the combination reads like a broken sentence, rewrite one. Second, avoid repeating the same phrase across assets, since redundant headlines waste slots the algorithm could have used to learn.

Should I pin headlines, and where?

Pinning is the escape hatch Google gives you to lock an asset to a specific position. It is also the most-abused control in the platform: it feels safe and quietly kneecaps performance.

Pinning reduces the number of combinations Google can test and generally lowers your Ad Strength, so use it surgically: pin your single strongest keyword headline to position 1, leave positions 2 and 3 open, and pin nothing else. That hybrid gives you guaranteed relevance up front while letting the system optimize everything behind it.

The mistake is pinning three headlines to position 1 and three to position 2 to force a fixed script. That collapses your 15-headline pool to the handful you pinned, the opposite of the variety the format rewards. If brand or compliance rules force you to control more than one slot, pin two or three genuinely different options to that position so Google can still test within it. Treat every pin as a testing slot you give up, and pin the minimum the business truly requires.

How do I know if the ad is any good before I spend?

Google gives you two live signals, and you should read both without confusing them.

Ad Strength rates the whole ad from Poor to Excellent based on variety, relevance, and asset quantity: more distinct headlines, more angles, more keyword coverage, higher strength. It is directional guidance, not a grade of copy quality, but it is worth chasing: Google reports that advertisers who lift Ad Strength on their responsive search ads and sitelinks from "Poor" to "Excellent" see about 15% more conversions on average. Push for "Good" or "Excellent" before launch.

Asset performance ratings show up after the ad has run and label each headline and description Learning, Low, Good, or Best. This is your live scoreboard. The playbook: keep "Best" assets and write new variations on the same angle, hold "Good" ones, and replace anything "Low" once it has enough impressions to judge (Google suggests waiting until at least 5,000 impressions before swapping a "Low" asset). "Learning" just means not enough data yet, so leave it alone.

This is what makes RSAs a compounding system rather than one-and-done. Each cycle you cut the "Low" lines and clone your "Best" ones, and the pool gets stronger. The operators who win at Google Ads are not better writers on day one; they just run this loop more times.

The honest limits

This framework will not save a broken offer or a weak landing page. RSAs improve which of your messages gets shown, not whether the deal is worth clicking. If your click rate is fine but nobody converts, the problem is downstream of the ad, on the page.

Two more cautions. Ad Strength is a guide, not gospel: an "Average" ad tied to a high-intent keyword can outperform an "Excellent" ad on a vague one, so let conversion data make the final call. And the ratings need real volume before they mean anything, so resist swapping headlines every few days on a hunch.

Prompt to generate a full 15/4 responsive search ad

This prompt builds the entire modular kit across all six buckets, enforces the character limits, and keeps every asset standalone. Paste it into your AI tool and fill the brackets.

You are a direct-response Google Ads copywriter. Write a complete
responsive search ad for this product:

PRODUCT: [PRODUCT OR SERVICE]
WHAT IT DOES: [ONE SENTENCE ON THE CORE OUTCOME]
TARGET SEARCHER: [WHO IS TYPING THE QUERY AND WHAT THEY WANT]
PRIMARY KEYWORD: [EXACT SEARCH TERM TO MATCH]
OFFER: [FREE TRIAL / DISCOUNT / GUARANTEE / PRICE]
REAL PROOF: [ONLY TRUE STATS OR CREDENTIALS, OR "none"]

Produce exactly 15 headlines (max 30 characters each, count
spaces) and 4 descriptions (max 90 characters each). Label each
headline with its bucket and distribute them as:
- 3 to 4 keyword-match (use the primary keyword)
- 3 to 4 primary-benefit (outcome / after-state)
- 2 to 3 offer/price
- 1 to 2 proof (ONLY if real proof was given; never invent numbers)
- 1 to 2 urgency (ONLY if the offer is genuinely time-bound)
- 1 to 2 direct-CTA

Give the 4 descriptions one job each: (1) benefit + offer,
(2) proof + a specific feature, (3) objection handling,
(4) CTA + soft deadline.

Rules:
- Every headline must read cleanly alone AND next to any other,
  because Google controls the order.
- No two assets may repeat the same phrase or idea.
- Cut filler like "your" and "our" to save characters.
- Do NOT fabricate any statistic, rating, or result.
- Show each headline's exact character count in parentheses.

Then name the single headline to pin to position 1, and why.

The pin recommendation at the end matters: it hands you the one surgical pin worth making and leaves the rest of the pool open. Launch with all 15, then start the cut-and-clone loop.

Still finding your footing on the platform? Our Google Ads for beginners guide covers campaign structure before you get to creative. For the craft underneath all six buckets, how to write a headline that converts is the deeper dive, and for the broader system of feeding AI into your ad workflow, start with how to use AI for ads.

Frequently Asked Questions

How many headlines should a responsive search ad have?

Provide all 15 headlines and all 4 descriptions. Google lets you run an ad with as few as 3 headlines and 2 descriptions, but Ad Strength drops toward "Poor" and you starve the algorithm of combinations to test. More distinct, non-repetitive headlines means more chances at the winning mix.

What is the character limit for responsive search ad headlines?

Each headline is capped at 30 characters and each description at 90 characters, with spaces counting toward the total. Optional display paths allow 15 characters each. Because 30 characters is only four to six words, cut filler like "your" and "our" so every character carries a benefit, keyword, or hook.

Should I pin headlines in responsive search ads?

Pin sparingly. Pinning locks an asset to a position but reduces the combinations Google can test and usually lowers Ad Strength. The best-balance approach: pin your single strongest keyword headline to position 1, leave positions 2 and 3 open, and pin nothing else. If you must control a slot, pin two or three different options there so testing continues.

What is a good Ad Strength for responsive search ads?

Aim for "Good" or "Excellent." Ad Strength rates variety, relevance, and asset quantity from Poor to Excellent, and Google reports that lifting it from "Poor" to "Excellent" on RSAs and sitelinks correlates with roughly 15% more conversions on average. Treat it as directional guidance, not a copy-quality grade, and let conversion data decide.

How do I improve a responsive search ad after it launches?

Use the asset performance ratings. Google labels each headline and description Learning, Low, Good, or Best once it has data. Keep and clone "Best" assets into new variations on the same angle, hold "Good" ones, and replace "Low" assets after they clear about 5,000 impressions. Run that loop repeatedly and the ad compounds.

Where to take this next

Responsive search ads reward operators who run the loop, not writers who nail one line. The framework here gets you a strong ad on day one; the ratings and a disciplined pinning strategy turn it into a machine that sharpens every cycle. Inside Asset Academy we share the exact prompt libraries and swipe files members use to fill all 15 slots fast and cut the losers. Join the Asset Academy community to get the full RSA prompt kit and swipe files.