Automated filtering of irrelevant search queries
The system takes keywords from Google Sheets and uses AI to check them against the rules for a specific product group. Queries are marked relevant or not, with a comment giving the reason. The solution cuts manual work and makes ad campaigns easier to scale.
- Industry
- E-commerce
- Geography
- Europe
- Period
- Running continuously
0
manual passes over lists100%
decisions with a reasonyou can see why a query was rejectedChannels
- Google Ads
- Google Sheets
- LLM
Context
Negative keywords are handled by hand: a specialist reviews thousands of queries, misses some, and budget keeps going to irrelevant traffic. Campaigns cannot scale with that volume of manual work.
What we did
- 01Relevance rules are described separately for each product group: different categories use different selection criteria
- 02Keywords are exported to Google Sheets and checked by the model against those rules
- 03Every query gets a relevant / not-relevant mark plus a comment explaining the decision
- 04The specialist reviews only the disputed cases and the rules themselves, not the whole list
Results
0
manual passes over lists100%
decisions with a rationale