Negative-keyword sweep

A negative-keyword sweep is the highest-leverage recurring cleanup in most Search accounts: find the searches that triggered your ads but should not have, and block them so the budget stops leaking.

Ask for it

Find wasted spend in the last 30 days and propose negative keywords.

What the assistant does

1. Groups the search terms by theme. analyze_search_terms sorts the last 30 days of search terms into negative-keyword candidates: people looking for information, jobs, something free, a competitor, and anything you told AdCopilot never to show for. It returns one ready-to-run add_negative_keywords call per theme, with the numbers behind it. Competitor terms are checked only if you have saved competitors in your business profile; AdCopilot never guesses one.

2. Checks the spend. analyze_waste names the zero-converting search terms and keywords. If it asks for raw rows instead, it runs a read-only search over the search-term view. The window is two YYYY-MM-DD dates (start_date / end_date), and AdCopilot builds the quoted segments.date filter:

   SELECT search_term_view.search_term, metrics.cost_micros,
          metrics.conversions, metrics.clicks
   FROM search_term_view
   WHERE segments.date BETWEEN '2026-08-17' AND '2026-09-15'
   ORDER BY metrics.cost_micros DESC
   LIMIT 50

3. Separates intent from noise. Not every zero-conversion term is waste; some are new or seasonal. Only the clearly off-intent ones become proposals.

4. Proposes the block with the exact terms, match type and level, and waits for your yes.

Choosing the level

Match type matters: a phrase-match negative blocks any search containing the phrase; exact blocks only that exact search. The assistant proposes a default and explains it; change it before you approve.