Replace weak ad components in responsive search ads

What it flagsComponents Google rates as Low

A responsive search ad consists of individual headlines and descriptions that Google combines itself. Once a component has had enough impressions, Google assigns it a performance label: Low, Good or Best (and Learning while there's still too little data). This recommendation looks at that verdict and finds ads where at least one headline or description carries the Low label.

You get a separate card per ad, with the ad group name included. That way you see exactly which ad has one or more weak components and how many there are.

Why it mattersWhy a single weak component holds your ad back

Google mixes and matches your headlines and descriptions per auction. A component with low asset performance gets shown less often and drags down the Ad Strength and CTR of the whole ad. Often it's not the whole ad that's underperforming, just a couple of weak components holding the rest back.

By replacing only those weak headlines or descriptions, you keep the proven components that are actually doing well. That's more targeted and less risky than rewriting an entire ad.

How it worksHow AdBuilder finds the weak components

AdBuilder looks at all active responsive search ads in active ad groups and campaigns over the last 30 days. There's deliberately no impression threshold and no cap on the number of ads checked: the gate is Google's own verdict. For each ad we pull the performance labels of every headline and description.

An ad becomes a candidate as soon as at least one component has the Low label. Only that explicit Low label counts as weak. Components without a verdict yet (Learning or unlabeled) are kept as is, since Google only rates those once there's enough data, and "no verdict" is no reason to discard something.

Candidates are sorted by CTR from low to high, so the ads lagging the most come first, and up to 5 ads are handled per run. The check runs weekly.

• Only responsive search ads that are active and have a landing URL
• Lookback window: 30 days
• Trigger: at least 1 headline or description with a Low label
• Only components with a Low label get replaced, the rest stay as is
• Maximum 5 ads per run, sorted by lowest CTR first

What you can doWhat you can do

AdBuilder pre-generates stronger variants for only the components labeled Low. The good and unproven components stay in place, and deliberately pinned headlines or descriptions keep their position. The proposal shows which components will be replaced and which are kept, along with their current CTR.

When applied, we build a new ad combining the kept and improved components. You choose whether it runs as an A/B test alongside the existing ad, or replaces the old one outright. You can still edit the copy before pushing it to Google.

The action is reversible: you revert it via History. Ads that already have an A/B test running are skipped, and after applying there's a short 14-day cooldown before the same ad is proposed again. If you'd rather leave an ad alone, you can dismiss the card.

• Start an A/B test (old ad stays live) or replace the old one
• Edit the copy of the new headlines and descriptions before pushing
• Apply now and revert later via History
• Dismiss the recommendation if you want to leave the ad alone