TLDR
Controlled tests across three real advertiser accounts found that Google's AI Max automated ad writing did not grow total revenue but shifted existing demand away from better-performing campaigns. The business-to-business account attracted more clicks but fewer sales because the system drew in consumer searchers who were never likely to buy. Nearly one in five automatically written ad text assets had to be removed before they received meaningful exposure. Advertisers should measure whole-account results, not just the new campaign's own dashboard, before deciding whether the feature is working.
KEY TAKEAWAYS
More activity, less money
Google's AI Max feature for search campaigns promised to save advertisers time by generating headlines, callouts and sitelinks automatically. Measured across whole accounts rather than just the new campaigns, the results told a different story.
Brad Geddes, co-founder of pay-per-click auditing platform Adalysis, ran controlled tests of AI Max's automated ad copy across three real advertiser accounts. AI Max took impressions, clicks and conversions from other campaigns while total account revenue fell.verifiedVerified Source: searchengineland.com[1] Geddes said the feature did not create new demand; it redistributed existing demand to less efficient channels.
"After further analysis, we found that AI Max was poaching impressions, clicks, and conversions from other campaigns and that the overall revenue for the account declined," Geddes said.[1]
How cannibalisation played out across account types
The three accounts covered different business types, and each showed its own version of the same underlying problem. In every case, the AI Max campaigns posted their own activity metrics while the accounts around them weakened.
An ecommerce advertiser selling more than 100,000 products saw AI Max cannibalise its carefully optimised and long-tail campaigns, pulling impressions and conversions away from the placements that had historically performed best.verifiedVerified Source: searchengineland.com[1] The scale of the catalogue made the cannibalisation harder to detect at first, because the losses were spread across thousands of product terms rather than concentrated in a single category.
Geddes said the pattern held regardless of account structure. The AI Max campaigns looked productive in isolation; the damage only became visible once total account revenue was measured against pre-test baselines.
Why the B2B account attracted the wrong audience
The business-to-business account produced the starkest result. The account's click-through rate jumped while its conversion rate dropped because the system pulled in consumer searchers who had no intention of making a business purchase.verifiedVerified Source: searchengineland.com[1]
Automated copy is built to attract attention broadly, not to filter for the specific buyer profile a B2B advertiser needs. Consumer searchers clicked through, found the product was not for them, and left. The advertiser paid for the traffic and received nothing in return.
Testing in the B2B account was stopped after three weeks when results proved poor and manual pinning of ad copy was restored.[1] Geddes said restoring manual control over the creative was the first step toward recovering the conversion rate the account had held before the experiment.
The ad copy quality problem that compounded the damage
Beyond the demand-shifting issue, the quality of automatically generated assets created a separate layer of risk. Approximately 19% of the automatically created ad text assets had to be removed before they received meaningful exposure.[1]
Geddes said, "One of AI Max's capabilities is creating assets for you."[1] That capability carries a cost: the advertiser must review every generated asset or risk copy that misrepresents the product, attracts the wrong searcher, or contradicts brand messaging going live in the auction.
For large accounts with dozens of ad groups, the review burden can exceed the time the automation was supposed to save. A 19% removal rate across the test accounts suggests AI Max's output required active editorial oversight, not passive acceptance.
What Australian advertisers should do before switching it on
Geddes's findings carry a direct implication for any Australian advertiser considering AI Max: the campaign-level dashboard is not the right measurement tool. A campaign can record strong clicks and apparent conversions while simultaneously depleting the performance of every other campaign in the account.
Adalysis's tests point to three practical steps before enabling automated ad copy. First, establish a clean whole-account revenue baseline over a period long enough to remove seasonal noise. Second, audit all automatically generated assets before they serve, not after. Third, for B2B accounts, restore manual pinning of copy as the default and treat AI Max's creative generation as an optional layer to test in tightly controlled conditions, not a feature to enable account-wide.
The tests were conducted across three accounts by Adalysis, with findings published on 28 July 2026.
SOURCES & CITATIONS
FREQUENTLY ASKED QUESTIONS
What is Google AI Max for search campaigns?
Why did total account revenue fall even when AI Max campaigns showed positive results?
Why was AI Max particularly damaging for the B2B account?
How many automatically written ad assets had to be removed in testing?
What should advertisers measure to assess AI Max fairly?
Jonas Valenti writes about search and how businesses get discovered. He has spent years watching what makes a company visible online, and is unsentimental about tactics that no longer work.
