Responsive Search Ad (RSA)

A Responsive Search Ad (RSA) is the main text-ad format for Google search campaigns. Instead of writing one fixed ad, you supply a pool of assets — multiple headlines and several descriptions — and Google's machine learning assembles them into ads, mixing and matching combinations and learning over time which perform best for each search. You provide a set of headlines and a set of descriptions, and when someone searches, Google selects a combination to show, factoring in the query, past performance, and other signals, displaying a few of your headlines and descriptions at a time. The format is named for being responsive: the ad adapts, showing more relevant combinations to different users rather than one static message. RSAs became the standard Google search-ad format after the older expanded text ads were retired, so for most advertisers building search ads today, the RSA is simply how search ads are made.

(Note that RSA, the advertising format, shares its letters with RSA the cryptography algorithm — an unrelated public-key system. This page is about the Responsive Search Ad.) RSAs matter because they change how search-ad creative is built and optimized. Rather than testing whole ads one against another by hand, you give Google a varied set of headlines and descriptions and let its system test combinations at scale, which can surface high-performing pairings a human might not try and which adapts the ad to the searcher. The trade is control: you no longer dictate exactly which words appear together, so the quality and variety of the assets you supply, and the structure you give them, become the levers. For a growth team running paid search, mastering RSAs means feeding the system strong, varied, on-message assets and using the controls it offers, because that is now how Google Ads search creative is assembled and optimized.

The cleanest contrast is with the expanded text ad (ETA), the format RSAs replaced as the search standard. With an ETA, you wrote a complete ad — fixed headlines and description in a set order — and tested whole ads against each other. The RSA flips this: you supply many headlines and descriptions, and Google's machine learning assembles and tests the combinations, choosing what to show per query. So ETAs gave the advertiser full control over wording and order but limited testing to whole-ad comparisons, while RSAs trade some of that control for automated, large-scale combination testing and per-user adaptation. ETAs are no longer created in Google Ads, which is why the RSA is now the default search-ad format. Knowing this history explains why search-ad best practice shifted from crafting individual ads to supplying strong, varied asset pools.

The RSA also sits among other Google formats, and it is worth keeping distinct from them. Responsive display ads apply a similar mix-and-test idea to the display network rather than search. Performance Max and other automated campaign types take the automation further, assembling ads across many channels. The RSA's specific niche is the search results page: text ads built from your headline and description assets, assembled by machine learning. The honest framing is that RSAs hand more of the assembly and testing to Google's system, which can improve performance through scale and adaptation but reduces granular control and demands disciplined asset quality. They are neither a magic upgrade nor a loss — they are a different model of building search ads, where the advertiser's job moves from writing finished ads to supplying and shaping the raw material the system works with.

Using RSAs well means feeding the system strong, varied, genuinely different headlines and descriptions, all on-message and relevant to the keywords and landing page, so the combinations it tests are all good ones. It means using the available controls thoughtfully — pinning a headline or description to a position when a specific message must always appear or for compliance, while leaving enough flexibility for the system to test freely — and watching the ad-strength guidance and performance data to refine the asset pool. It means writing assets that work in different combinations, since they will be mixed, and aligning the whole ad with the landing page and offer. Done this way, RSAs let Google's machine learning do the combination testing at scale while you control quality and message, which is the right division of labor for the format.

The failures are supplying weak, repetitive, or off-message assets (so every combination is mediocre), over-pinning until the format can no longer test and the benefit is lost, writing headlines that clash when combined, and treating the RSA as fully hands-off rather than shaping and refining the asset pool. There is also the simple confusion of mistaking this RSA for the cryptography algorithm of the same letters. The discipline is to give the system excellent, varied, compatible assets, use pinning sparingly and deliberately, align the ad with the landing page, and keep refining — so the Responsive Search Ad's machine-learning assembly works with strong raw material rather than papering over weak creative.

A Responsive Search Ad (RSA) — Google Ads' standard search format built from multiple headlines and descriptions assembled by machine learning — is distinct from RSA the unrelated cryptography algorithm.

Etymology: source.

Search interest for this term over the last five years:

View interest-over-time on Google Trends →

Areas of marketing where responsive search ad (rsa) is a core concern: