Search Terms
Search Terms shows the actual queries users typed before clicking a Google Ads ad, with cost/conversion/ROAS per query. It’s the canonical surface for negative-keyword mining.
What you get#
- Query-level metrics — cost, impressions, clicks, conversions, CPA, ROAS for each synced search term.
- Match-type column — exact / phrase / broad / close-variant, so you can spot low-quality variants coming in through broad-match experiments.
- Campaign + ad-group context — which campaign and ad group the query matched into.
Common flows#
- Negative-keyword candidates. Filter by spend > $X and conversions = 0. Export as a negative-keyword list.
- Match-type leakage. Group by match type; if close-variants or broad terms dominate the long tail with poor ROAS, tighten the match types on the originating ad group.
- New keyword candidates. Filter by high CTR + high conversion rate on queries you don’t currently bid on. Add them as new keywords in the originating ad group.
Platform coverage#
Search-term data is Google Ads only. Meta and TikTok don’t expose equivalent query-level data.
From MCP#
Search-term data isn’t directly exposed via MCP today. To analyze search terms from an external agent, generate a report (adlyse-reports.generate_report) — the optimization_activity_log and monthly_performance templates surface search-term analyses inline.