The 30-day search-term waste audit

A repeatable method for finding the spend your search campaigns leak, and shipping the negatives that stop it.

PlaybookBY THE ASTROFABRIC TEAM · AUG 2, 2026 · 4 MIN READ

Every search account leaks: queries that match your keywords and will never match your product. The fix is unglamorous and compounds beautifully - a monthly audit that turns the search-term report into negative keyword lists. Here is the method as an agent runs it.

Step 1: pull the terms

Last 30 days of search terms with spend, clicks and conversions, per campaign. Sort by spend with zero conversions - the head of that list usually pays for the whole exercise.

Step 2: classify with rules first

Bucket terms as convert, plausible, and mismatch. Deterministic rules catch most mismatches - job-seeker language, free-seeker language, wrong geography, competitor-only intent you do not bid on. Judgment (human or model) handles the plausible middle; the data handles the rest.

Step 3: build the negative sets

Turn mismatch clusters into negatives at the right level: account-wide for universal junk, campaign-level for positioning conflicts. Prefer phrase negatives for patterns and exact for surgical cuts, and keep the lists named by theme so next month's pass extends instead of duplicating.

Step 4: ship and schedule

Land the analysis where the team works - the full table in a sheet, the top fixes in the channel - and apply the negatives through your normal approval path. Then schedule the whole loop monthly. On AstroFabric this is one standing mission with the analysis delivered to Sheets and Slack, and changes staged for a human to approve.

⟨ RUN IT INSTEAD OF READING IT ⟩

Every playbook on this blog ships as a runnable mission.

Open a free workspace and the playbook library is waiting - describe the outcome and the agents carry it end to end, with 1,000 trial credits on us.

⟨ KEEP READING ⟩
ArticleAgentic marketing

What is an AI marketing agent?

A working definition, the difference from automation, and the checklist that separates a real agent from a chatbot with opinions.

Aug 8, 2026 · 5 min read