Which Google Ads prompts have we tested across AI setups?

What does each setup score, job by job?

At a glance, by Google Ads job. Open a job to see the method, or a pack for a single creator's prompts with the real output. Rows marked not yet tested have no run behind them; we never invent a result to fill a cell.

Every measured cell traces to the run on its linked page, where the real output is shown. Rows without a run stay marked "not yet tested" until one publishes; this scoreboard invents nothing.

Why is "tie" a good result, not a loss?

Some of this work is a straight read of your own account: sorting search terms into buckets, or reading conversions and ROAS off the search terms report. Any competent setup gets those right, so a tie there is the result we expect, and we say so plainly. AdTAO earns its keep on the checks a single-account tool cannot make: before it hands you a negative keyword, it reads the negatives your account already blocks so it will not have you re-add duplicates, and it checks the term against results from accounts like yours so it will not block one that earns money elsewhere. That is the difference you are paying for, and it is the difference the wins below measure.

Open a job or a creator pack

What should I do with this?

Start with the search-term pack: it turns "clean up my search terms" into a ready-to-paste negative keyword list that skips the duplicates your account already blocks, so you paste the final list straight into a shared negative keyword list in the Google Ads UI. If you let your AI agent make changes for you, read the approve-before-change job first, so a wrong instruction lands in your approvals queue instead of live in the account. Both are a few minutes to act on, and every negative keyword traces back to a row you can filter to in your own search terms report by setting conversions to 0.