On September 1, 2026, Google began moving Search campaigns built on campaign level broad match or Automatically Created Assets into AI Max, without asking the advertiser first. Most accounts have no opt-out. By September 30 the switch will be complete industry wide.

This is not a feature launch. It is a default, and defaults are never neutral, they are choices made by someone else and dressed up as inertia. The advertiser who does nothing on September 1 has, in effect, delegated a live budget decision to Google's own optimization layer, whether or not they ever clicked accept.

What Actually Changed on September 1?

Google auto migrated any Search campaign still running on campaign level broad match or Automatically Created Assets into AI Max, a bundle that layers AI generated search terms, expanded creative combinations and broader matching on top of existing keywords, according to the migration timeline reported by Search Engine Land. The rollout is staged through the month, so an account can look untouched on September 2 and be fully migrated by September 25, with no plain language email marking the moment it happened.

The account owner still sees a dashboard with keywords and match types on it. What the dashboard does not show clearly is that the matching logic underneath has already been swapped for a version the advertiser never tested.

What Does Google Promise, and What Do Independent Tests Find?

Google's own number for AI Max has already moved once. At the feature's May 2025 launch, Google cited a 14 percent lift in conversions or conversion value; by the April 2026 announcement retiring the older campaign types, that figure had been revised down to 7 percent, and the 7 percent figure explicitly excludes retail advertisers, per reporting on the migration by PPC Land.

Retail is exactly where the independent numbers get worse. Smarter Ecommerce analyzed more than 250 retail campaigns and found a median revenue lift of 13 percent, close enough to Google's own claim, but median cost per acquisition rose 16 percent in the same accounts, and ROAS across the sample ranged from plus 42 percent to minus 35 percent depending on the account. Monks agency ran a separate test across roughly 30,000 AI Max search terms and found that 99 percent of the resulting impressions produced zero conversions.

None of this makes AI Max worthless. It makes it variable in a way the headline number hides, and variability is the wrong thing to inherit as a default with no easy way back.

Why Isn't Anyone Refusing the Default?

Because refusing a default takes effort, and avoiding that effort is the entire design purpose of a default. Behavioral economists call this status quo bias: once an option is pre selected, people keep it not because they evaluated the alternative and rejected it, but because switching costs more attention than staying put. The clearest evidence for this comes from organ donation policy, not advertising: countries that set donation as the default enrol dramatically more donors than countries that require an active opt-in, and the gap has nothing to do with how people actually feel about donating.

Google Ads is now running the same mechanism on media budgets. The advertiser did not evaluate AI Max and decide the tradeoff was worth it. The advertiser simply did not click the one setting that would have stopped the migration in time, usually because they did not know the setting existed.

A default is not a recommendation. It is a decision made in advance, on your behalf, by someone who does not answer for the outcome.

What Is This Actually Costing the Budget?

It costs visibility first and money second. The State of PPC 2026 survey, run by PPCsurvey.com across 1,306 paid search professionals, the largest sample in the survey's history, found that 53 percent now say managing paid media is harder than it was two years ago. 62 percent point to increasingly black box platforms as the reason, ahead of measurement problems and rising competition. Only 16 percent say it has gotten easier.

Put the two data sets together and the shape of the problem is clear. Accounts are being defaulted into a system with a documented 16 percent CPA increase in retail, at the same moment the people running those accounts report losing the visibility they would need to catch it early. The cost is not that AI Max always underperforms, several accounts in the same Smarter Ecommerce sample gained ROAS. The cost is that nobody chose to accept that variance. It was assigned.

So What Does a Brand Director Actually Do This Month?

Check whether the decision has already been made, then decide deliberately instead of by default.

• Pull the campaign settings page and confirm which Search campaigns have already migrated to AI Max versus which are still pending through September 30
• Segment the last 30 days of spend by AI Max versus non AI Max campaigns before comparing this month to last, a blended average will hide a CPA spike inside a revenue gain
• Ask the agency or in house team for the search terms report specifically, not the conversion summary, that is where the Monks style zero conversion waste actually shows up
• Set a CPA and ROAS alert threshold now, before the migration completes on September 30, not after the first invoice looks wrong

None of this requires distrusting automation on principle. It requires treating a platform default the way any other unreviewed budget line deserves to be treated: checked before it compounds across a full quarter.