Responsive Search Ads Audit: 10 Checks Beyond Ad Strength
Responsive Search Ads Audit: 10 Checks Beyond Ad Strength
Ad Strength is the first thing most people look at when reviewing responsive search ads, and it is close to the least useful. It measures whether you have given Google what Google asked for: lots of assets, keywords in headlines, nothing pinned. It does not measure whether the ad sells anything.
Accounts full of Excellent-rated ads that convert poorly, and Good-rated ads that quietly outperform them, are common enough that the label should never be the audit. This is the audit: ten checks that look at what your RSAs actually do, not what the interface thinks of them.
1. Asset performance labels, read properly
Open the asset report and look at the Low, Good and Best labels, but read them with their limitation in mind: they reflect impression-weighted engagement, not conversion. A Low label on a high-volume ad is worth acting on. A Low label on an asset that has barely served is noise. Replace genuine Low performers with new angles rather than rewording the same claim, because a synonym of a weak message is still a weak message.
2. Whether Google can actually see your best message
Count your distinct selling propositions across the headline set. Many RSAs contain twelve headlines that say four things. Delivery speed three ways, price two ways, brand name three ways, and a couple of generic fillers. The machine can only test the messages you give it. If a proposition that matters to buyers, guarantees, stock, expertise, compatibility, is not in the asset pool at all, no amount of optimisation will surface it.
3. Pinning: audit the reason, not the rule
Pinning gets treated as either forbidden or mandatory, and both positions are lazy. The audit question is whether each pin has a reason that survives scrutiny. Compliance text that must appear, a brand rule about position one, a proven winner protected in place. Those are legitimate. Pins that exist because someone wanted the ad to read nicely in the preview are throttling combinations for no benefit. Check the combinations report for what pinning has done to serving variety.
4. Headline and description truncation in the wild
Preview tools show the ad you hope for. The combinations report and real SERP checks show the ad people get. Look for headlines that only make sense in a specific order, descriptions that get cut mid-claim on mobile, and combinations where two headlines repeat the same phrase side by side. Every asset should be written to stand alone in any slot, because that is how it will serve.
5. Keyword and query alignment per ad group
For each significant ad group, put the top converting search terms next to the RSA assets. The question is simple: does the ad speak the language of the queries that pay for it? Ad groups drift. Products change names, new query patterns emerge, and the RSA written eighteen months ago is answering questions people have stopped asking. Where the top terms and the headlines share no vocabulary, rewrite.
6. One RSA per ad group, used properly
Google recommends a single RSA per ad group, and most accounts comply. The audit point is what that means for testing: your asset pool is the test. Check that each ad group's RSA actually uses its slots, fifteen headlines, four descriptions, rather than the minimum, and that new assets have been added within a sensible window. An RSA untouched for a year is not a mature ad, it is an abandoned test.
7. What AI-generated assets have been doing
If automatically created assets are enabled, audit what they have written on your behalf. Pull the asset report, filter to automatically created, and read them as a compliance and brand exercise. Some will be fine. Some will make claims you would never approve. Either curate them actively or turn the feature off, but do not let it run unread.
8. Ad copy versus landing page promise
Take the strongest claims in the asset pool and check each one against the landing page it serves to. Free delivery thresholds that changed, sale language outliving the sale, guarantees the page no longer mentions. Mismatches here do not just waste the click, they train the bid strategy on conversions from misled visitors, which is worse than no data.
9. Coverage of the buying objections
Good RSA asset pools answer the reasons people do not buy, not just the reasons they might. Price objection, trust objection, urgency, comparison with the obvious alternative. Audit the pool against a simple objection list for the product. Most pools are eight variations of the value proposition and nothing that handles doubt.
10. Performance dispersion across ad groups
Finally, zoom out. Rank ad groups by conversion rate and look at the spread. A wide spread with similar traffic quality usually means the copy work has been uneven: someone wrote the top ad groups carefully and templated the rest. The template ad groups are where the next win is, because they have proven traffic and unproven copy.
The habit that keeps RSAs healthy
None of these checks is sophisticated. What separates accounts is cadence. A quarterly pass through the asset report, replacing genuine Low performers, refreshing one message angle, and re-reading the automated output takes an hour per account and compounds.
Ad Strength will not tell you any of this. The asset report, the combinations report, the search terms and the landing pages will. That is where the audit lives.