Google retired Expanded Text Ads in June 2022. Every new search ad since then has been a Responsive Search Ad. RSAs give Google 15 headline slots (up to 30 characters each) and 4 description slots (up to 90 characters each) to mix and match based on the search query, the user, and the competitive context. Google selects the combination it predicts will win the auction.
This changes what good ad writing looks like. With Expanded Text Ads, you wrote three versions of a full ad and chose which to run. With RSAs, you write 15 independent headlines that Google combines with each other in any order. The headlines must make sense as standalone units and in any two-headline combination Google might assemble. Most ad copy written for the Expanded Text Ad era does not meet this requirement.
What 'Ad Strength' actually measures is the most misunderstood metric in RSA optimization. Google's Ad Strength rating — Poor, Average, Good, Excellent — measures the diversity and relevance of the assets you have provided, not the actual performance of the ad. An ad with 'Average' Ad Strength can outperform an 'Excellent' rated ad. Ad Strength rewards quantity of unique headlines and keyword coverage; it does not predict CTR, Quality Score, or conversion rate. Use Ad Strength as a completeness check, not a performance predictor.
The headline writing framework that produces the best RSA inputs: write three types of headlines. First, keyword-mirroring headlines that match what the user searched — if your keyword is 'project management software,' one headline should be 'Project Management Software.' Second, benefit headlines that state the value proposition — 'Save 5 Hours Per Week,' 'No Per-User Pricing.' Third, CTA headlines — 'Start Free Trial,' 'See Plans & Pricing,' 'Get a Demo.' This three-type mix gives Google combinatorial flexibility to assemble ads that match both the search query and the stage of the user's consideration process.
The headline pinning feature is widely overused and frequently counterproductive. Pinning forces a specific headline into Position 1, 2, or 3 of the ad. Every headline you pin is a headline Google cannot move to where it would perform best. Pinning Position 1 to your brand name prevents Google from testing whether a benefit headline in Position 1 generates higher CTR than the brand name for certain query types. The right use of pinning: pin one headline if regulatory or legal requirements demand a specific claim appears in every ad. Do not pin for preference.
Headline length and specificity matter. Google's matching uses your headlines to determine ad relevance for the auction. Vague headlines — 'The Best Solution,' 'Try It Today,' 'Learn More' — contribute almost nothing to relevance scoring. They will be selected less often, dragging down your combination pool. Specific, descriptive headlines that contain actual keywords, actual numbers, or actual product terms contribute to relevance and get served more frequently.
Descriptions have different rules than headlines. Descriptions are 90 characters and appear beneath the headlines. Google selects 2 out of your 4 descriptions for each impression. Unlike headlines, descriptions can and should contain complete selling thoughts with clear beginning and end. Use description slots for proof points that need more than 30 characters: 'Used by 12,000 teams — G2 top-rated for three years running' or 'Full-funnel reporting from first click to closed revenue — connect in 5 minutes.'
Performance Max and RSAs compete for the same queries. If you run both Search campaigns with RSAs and Performance Max campaigns, the two campaign types can compete against each other in auctions. Google gives Performance Max priority in most cases when both are eligible. This can suppress RSA impression share in ways that look like poor creative performance when the actual issue is campaign type competition. Monitor search impression share loss data at the campaign level to distinguish poor creative from campaign priority overlap.
The Quality Score components an RSA can improve: RSA directly influences Expected CTR and Ad Relevance — two of the three Quality Score components. Landing Page Experience is set by your destination URL, not the ad. To improve Expected CTR, test headline combinations that match high-volume queries precisely. To improve Ad Relevance, ensure your top-priority keywords appear verbatim in at least 2–3 headlines.
Creative testing with RSAs requires a different methodology than ETA testing. You cannot run a clean A/B test between two RSAs using the same asset pool — combination selection makes the test uncontrolled. The practical testing approach: run two RSAs in the same ad group with entirely different headline pools (one pool focused on benefits, one focused on features) over a statistically significant period. Asset performance reports show which headline type gets served more and generates better CTR. Digital Face analyzes your RSA Asset Strength, headline coverage distribution, and combination performance data, surfacing underperforming headlines and recommendations for new copy directions based on your search term data. Free plan at digital-face.nl, no credit card required.