gt-google-ads

Start with conversion tracking, then review account structure, negative keywords, bidding, Quality Score, and PMax in that priority order. Pull 90 days of search terms sorted by cost, flag queries with $50+ spend and zero conversions, and finish with a dollar-figure wasted-spend tally and a ranked fix list.

Set up Enhanced Conversions for Leads with hashed first-party data, capture GCLID on forms, and import offline conversions from HubSpot or Salesforce as lifecycle or opportunity stage changes with value. This lets value-based bidding optimize toward SQLs and revenue rather than raw lead volume.

Only after Search campaigns are clean and conversion tracking is solid. Cap PMax at 15-25% of total spend, apply account-level negative keywords and brand exclusions, theme asset groups by intent, and judge results on incremental pipeline rather than platform-reported ROAS.

Use exact match for proven high-intent BOFU terms, phrase match for controlled expansion, and broad match only with Enhanced Conversions, value-based bidding, and a maintained negative list. Test broad match or AI Max through the experiment tool before enabling it on live campaigns.

Yes, but it is read-only by default. Live analysis uses GAQL query templates through the API or an MCP connector, and any mutation such as pausing campaigns or changing bids follows a Confirm-Execute-Post-check protocol requiring explicit user approval for each scoped change.

Smart Bidding needs roughly 30 or more conversions per month and clean signal to work. B2B accounts often feed it page-load conversions or junk form-fills, so the algorithm optimizes toward cheap leads; fixing tracking and importing offline conversions is the prerequisite.