Search Terms Report: The Most Underread Ads Report

The search terms report is the Google Ads report showing the actual queries that triggered your ads and drew clicks — not the keywords you bid on, but the searches real people typed. The two lists agree less than intuition expects, and every interesting fact about the account lives in their gap: keywords are what you asked for; search terms are what you paid for. Most wasted spend is visible nowhere else. And many of the best keyword ideas you will ever have are already in it, written by your own customers.

How the search terms report works

The report exists at campaign, ad group and keyword level in the UI, and through the API as search_term_view — the query text, the keyword and match type that caught it, and full performance columns: impressions, clicks, cost, conversions.

What it does not show is everything. Google's privacy thresholds withhold queries searched by too few users, so part of your click volume maps to unnamed terms rolled into an "other" bucket. Two honest implications: the visible report understates the long tail, and the hidden spend can only be steered indirectly — through match-type radius and negatives that generalise. That makes the visible portion more precious, not less. It is the sample you can act on.

Reading it: waste, harvest, drift

Three reading modes, three different profits:

• Waste. Terms with spend and no conversions — especially terms whose intent was never yours: free-seekers, job-hunters, DIY researchers, someone else's spare parts. The exit is negative keywords, added at the right level.
• Harvest. Converting terms not yet in your keyword list — customers phrasing your product better than you did. The exit is promotion to keywords, often exact, with copy to match.
• Drift. The slow divergence between what a keyword means to you and what it matches — visible only across time, keyword by keyword. The exit is match-type tightening, or negatives fencing the boundary.

A single pass always finds some waste and rarely shows drift. The report pays by being read on a cadence, not once.

Reading all of it, finally

The search terms report was always the highest-yield, least-read report in Google Ads, and it was underread for one reason: volume. Thousands of rows, most of them fine, and attention as the scarce input. That constraint is the one AI removes.

With account access, your agent reads the whole report and the work becomes conversational: "search terms from the last 30 days with spend and no conversions, worst first"; "which new terms appeared this week that no keyword covers?"; "cluster the losers — what do they share?" Pattern-finding across the whole table is what n-gram analysis formalises: the single word carrying spend across hundreds of losing queries that no row-by-row skim would ever surface.

And the loop closes in the same conversation. Identified waste becomes a negative-keyword call your agent stages and executes on your word — you stay in control, every change logged; the harvest becomes drafted keywords; next week's pull shows whether the fence held. The weekly ritual — new terms, waste, harvest, negatives, done — drops from an hour of table-scrolling to minutes of review. That is the real shift of the AI era: the report did not change. The cost of actually reading it did.