Start free
Features › Google Ads Reporting & Insights › responsive search ad headline performance

Responsive search ad headline performance, asset by asset — and a reviewed way to replace the ones that don't work.

Google reports responsive search ads as a whole and their assets individually, and the asset view is where the useful signal is: which headlines are actually being served, and what happens when they are. AdCortex aggregates asset-level metrics into a headline report, finds the words that work across headlines, audits pinning, and stages replacements.

Last verified: September 10, 2026 App version: 50.1 Canonical page for this topic: Reporting & Insights

Reading the table

Sort by impressions to see which headlines Google actually chooses to serve — often a third of the headlines in an RSA carry most of the impressions. Sort by conversions or conversion rate to see which of the served ones work. The gap is the list: headlines Google serves that do not convert are candidates for replacement; headlines that convert but are rarely served are candidates for pinning or for a copy variant that Google will pick. Delta badges compare with the prior period so a headline that was fading is visible before it is dead.

Campaign breakdown and the ad-group edit mode

Expanding a headline shows its performance per campaign. An ad-group edit mode narrows the view to one ad group's ads so the replace flow can be scoped precisely. The point of the breakdown is to stop a global decision being made on a blended number: "Free Shipping" at a 4% conversion rate overall may be 7% in the retail campaign and 1% in the B2B one, where it should not be running at all.

N-grams over copy

The n-gram table tokenises headline text and aggregates performance by word and phrase across all headlines containing it. It answers a different question from the headline table: not which headline, but which idea. "Same-Day" across eleven headlines at a conversion rate 40% above the account's is a finding that survives any single headline being retired, and it is the input that makes the next Headline Creator run better — the account's top-performing phrases are what it drafts from.

The pinning audit

Pinning a headline to position 1 forces Google to show it there in every combination, which is sometimes required — a brand name, a legal line — and often the reason an RSA underperforms. The audit lists pinned headlines by position and compares the performance of pinned and unpinned ads, so a pin added a year ago for a promotion that ended is found rather than inherited.

Replacing a headline, reviewed

Pick a headline, open the update modal, and the report lists every ad containing it with the campaign and ad group. Tick the ads — all of them, or only the prospecting ones — enter the replacement, and stage. Each ad becomes a Checkout row with the before and after text; the push updates the ads. Google treats an edited RSA as a new ad for learning purposes, which is why the row exists: a person should decide that a headline change across thirty ads is worth resetting their history.

  • Worked example: "Call Now For A Quote" serves in 14 ads, converts at 1.1% against an account rate of 3.4%, and is pinned to position 2 in nine of them. The n-gram table shows "Free Quote" at 5.2% across six other headlines. Replace with "Get A Free Quote Today" in the 12 prospecting ads, leave the two brand ads, and unpin in Google Ads. Checkout shows 12 ad-copy rows with before and after.

The loop with Headline Creator

This report and the Headline Creator are two halves of one cycle. The report finds what works — the n-grams and headlines with conversion rates above the account's — and what does not. The Headline Creator drafts new RSA copy for an ad group from that same performance context and the account's brand terms, creates the ad paused, and stages it; activation is a separate reviewed step because that is when it starts spending. Two weeks later the report shows whether the new headlines are being served and converting, and the cycle repeats. The Ad Competitor does the same with a challenger framed against the ad group's incumbent ads.

What asset-level reporting can and cannot tell you

Google attributes an impression, click, or conversion to every asset that was in the served combination, so a headline's numbers describe the combinations it appeared in, not the headline alone. Two consequences. A headline that only ever serves alongside a strong description inherits its conversion rate; the n-gram view partly corrects for this by aggregating across many combinations. And a headline Google rarely serves has too little data to judge — low impressions is a finding about Google's choice, not the headline's quality. Treat the report as directional and let the pinning audit and the campaign breakdown supply the context before replacing anything.