Responsive search ads and how to test them

What are responsive search ads?#

Responsive search ads are the default text ad format in Google Search campaigns, and the shape of the format is a fox den of interchangeable parts: you author the pieces, and the serving system assembles the finished ad for each query. One ad accepts up to 15 headlines and 4 descriptions, per Google Ads Help, 2026, with every headline capped at 30 characters and every description capped at 90 characters.

Per impression, Google serves up to 3 headlines and 2 descriptions, and the later slots are never guaranteed to appear, so every asset has to make sense standing alone. That constraint is exactly why the dynamic search ads vs responsive search ads question keeps surfacing in planning meetings: one format hands the copy to a machine that reads your website, the other hands it to you.

Expanded text ads are gone. Since 30 June 2022, responsive search ads have been the only Search ad type you can create or edit in a standard Search campaign, per Google Ads announcement, 2022, a fact still true when we checked on 11 August 2026. The practical consequence carries the whole entry: the unit of test is the asset, not the ad. You no longer split test ad A against ad B; you feed the machine distinct parts and judge each part on its own trail of data. For how these limits behave outside English, folkfox's character limits and double byte truncation entry covers the truncation edge cases.

A responsive search ads example#

A responsive search ads example for a Malta ecommerce brand might seed headlines such as “Free EU Delivery Over €50” (an offer), “Softer Sleep In 30 Nights” (a benefit), “Rated 4.8 By 2,100 Buyers” (social proof) and “Malta’s Certified Sleep Studio” (authority). Four headline hooks, four genuinely distinct directions, and the serving system can pair any of them with any description. That is the sharp specification in action, and the table below pins the responsive search ad character limits down precisely.

Dynamic search ads vs responsive search ads#

The dynamic search ads vs responsive search ads question is really a question about authorship. A dynamic search ad generates its headline and landing page match from your website content, which makes it a cunning safety net for large catalogues but a poor place to test copy, because nobody on your team wrote the words being tested. A responsive search ad flips that arrangement: you write every asset, and the machine only chooses which combination to show. Both formats live inside Search campaigns, the top level structure described in Google Ads API docs, 2026, but only one of them hands you a controlled creative surface to run experiments on.

The table below sets the dynamic search ads vs responsive search ads distinction out plainly, stripped of jargon. Read it as a decision aid rather than a verdict: DSAs still earn their keep on sprawling catalogues where nobody has time to write ten thousand headlines by hand, but for any product, service or landing page you actually want to test creative on, the responsive search ad format is where the culling loop below can go to work, patiently pruning the pack until only proven performers remain.

For the wider campaign picture, the folkfox search campaigns guide covers structure, match types and bidding, and the folkfox Google Ads hub holds the rest of the encyclopedia. This entry stays on the creative trail: the specification above, and the seeding and culling below.

In practice, most accounts settle the dynamic search ads vs responsive search ads question inside the first ad group they build. A retailer with forty near identical product pages might keep a handful of DSAs as a safety net, while every campaign built around a real offer or a named benefit runs on responsive search ads from day one. The two formats are not rivals fighting for the same budget line; they are tools for two different jobs, and confusing them is the quickest way to waste a testing quarter.

Seeding, and how many RSAs per ad group#

Once the dynamic search ads vs responsive search ads decision lands on responsive search ads, seeding is the next job, and how many RSAs per ad group is the first practical limit you hit. The ceiling is a hard cap of 3 enabled responsive search ads per ad group, per Google Ads account limits, checked 11 August 2026, and in practice one or two well seeded ads per ad group is plenty. Spreading thin copy across three ads just dilutes the data; a steady seeding of one strong ad usually beats a scattered spread of three weak ones.

Seed each ad with 10 to 15 distinct headlines and 3 to 4 descriptions, and make them genuinely different hooks rather than paraphrases: an offer, a benefit, social proof, authority. Google’s own guidance in Google Ads best practices, 2026 pushes the same way, asking for headlines that highlight different things and warning against redundant phrasing. Fifteen versions of “Best Prices Today” teach the system nothing; the algorithm can only find a winner if the candidates carry genuinely distinct directions.

Pinning deserves a wary word. Pinning locks an asset to one position, which is occasionally a compliance necessity, but every pin shrinks the combination space the machine can prowl. If legal copy must always show, pin two or three different approved lines to the same position rather than one, so the serving system keeps some room to test. Otherwise, leave the den door open and let the combinations run. Our sibling entry on pinning headlines and descriptions goes deeper on the multivariate cost of a single pin.

The RSA asset report and the culling loop#

The RSA asset report is where dynamic search ads vs responsive search ads theory turns into weekly practice. Open any responsive search ad and view asset details to see each asset listed with its own row, per Google Ads asset report guide, 2026. Historically the report labelled assets Low, Good or Best once they left a Learning state, and the trap practitioners fall into is treating those labels as verdicts.

They are relative rankings inside one ad, not significance tests: Best simply means best of this particular litter, and a Low asset sitting in a strong ad can still outperform a Best asset sitting in a weak one. Checked 11 August 2026, Google notes that this performance column has been superseded by full performance statistics for dates on or after 5 June 2025, so expect per asset impressions, clicks and conversions alongside or instead of the old labels.

Ad Strength is the other rating on the tin, and it needs the same measured moves. It scores the quality and diversity of what you loaded before serving ever begins, per Google Ads Ad Strength guide, 2026, which makes it a predictive checklist, not a practical proof of performance. Chase it to Good, then let real conversion data outrank the checklist. Padding an ad with filler copy purely to reach Excellent is creative theatre, and the culling loop below is the antidote.

That loop, cull the Low, replace with a fresh concept, change one thing at a time, is the whole method behind any serious dynamic search ads vs responsive search ads programme once RSAs have won the argument. It is slower than the old A/B ritual and far more honest about what the machine actually rewards. Run it monthly on your money ad groups and the creative culling compounds; run it never and your budget quietly funds the same tired lines while the asset report keeps track of exactly how much. If you would rather a fox ran the loop for you, our PPC management service does exactly this.

Deadlines and disapproval traps#

Two dated facts belong in every RSA plan, both checked 11 August 2026. First, call only ads are being retired into responsive search ads plus call assets: all options to create a new call only ad were removed in February 2026, and the full sunset lands in February 2027, when existing call only ads stop receiving impressions, per Google Ads transition notice, 2026. If phone calls feed your funnel, the migration is not optional; rebuild the call experience as call assets on an RSA before the deadline does it for you.

Second, the Editor trap. Google Ads Editor validates the 30 character limit at upload, so a headline that fits will post cleanly, but Editor is blind to policy. The policy review only begins after your changes post, per Google Ads review process, 2026, which means a clean upload can still disapprove hours later for trademark use, punctuation, capitalisation or restricted claims. The silent suppression pattern is worse: the upload succeeds, the disapproval email lands overnight, and the ad group sits dark until someone checks. After every bulk upload, follow the scent back into the web interface and confirm approval status before you log off.

These two deadlines sit outside the dynamic search ads vs responsive search ads decision entirely. They apply to your account regardless of which format writes the headline, so put them on the same calendar reminder as your next culling pass.

Build a standing check around both deadlines rather than trusting memory. A quarterly reminder that asks two questions, are any call only ads still live, and did the last bulk upload actually clear policy review, catches most of the damage before a client notices a quiet ad group. The same reminder is a natural moment to run the culling loop above, so the admin housekeeping and the creative housekeeping happen together instead of competing for attention.

Frequently asked questions#