What 23 tests reveal about Google AI Max performance

Source
google-adsai-maxbroad-matchsearch-term-matchingtext-customizationurl-optimizationroasquality-scoreperformance-maxsearch-bidding-exploration

Summary

9-month study of 23 AI Max tests across 16 mature advertisers shows that enabling all three AI Max features simultaneously (search term matching, text customization, URL optimization) delivers 40% higher success rates than search term matching alone. Account-wide rollout is critical - over half of “new” queries at campaign level were already captured elsewhere in the account.

Key Insight

  • All-in beats incremental: campaigns using all 3 AI Max features saw 40% higher uplift vs baseline search term matching only. Yet only 50% of tests enabled text customization and just 44% enabled URL optimization - most advertisers are leaving performance on the table.
  • Text customization lifts Quality Score: average QS improved from 6.8 to 7.3, with ad relevance showing the biggest gain. Headlines were modified far more often than descriptions.
  • Account-level view is essential: at campaign level, AI Max showed +7% conversion value from never-targeted queries. But at account level, 54% of those queries were already captured by other campaigns - real incremental uplift is ~3% for accounts already running high broad match.
  • DSA conflict: every successful AI Max test occurred in accounts with low or no DSA adoption. DSA and AI Max likely compete for the same inventory.
  • Brand inclusion lists are now AI Max-exclusive - another reason to adopt.
  • Search Bidding Exploration (SBE) is a natural companion to AI Max, and early results suggest hyper-consolidation of account structure may work.
  • Don’t judge by match type CPA - the query reshuffling between campaigns makes campaign-level CPA by match type misleading. Full account impact is the only valid measure.