Google AI Max is about to turn a familiar paid search setting into a broader operating decision. On September 1, campaigns using Google's campaign-level broad match setting or automatically created assets will begin moving into AI Max for Search campaigns, according to an advertiser notice reported by PPC Land.
That date is close enough to matter and far enough away for teams to pretend they have time. They don't have much. The risk isn't that Google is adding another AI feature to the interface. The risk is that query expansion, creative generation, and campaign control start arriving as one package, while the account owner thinks they only opted into one of them.
AI Max changes the bargain
Google's explanation is designed to sound reassuring. AI Max is supposed to preserve the intent of the older setup while bringing more automated matching and asset decisions into the campaign. For marketers, the important word is not preserve. It's inherit.
Campaigns with automatically created assets enabled will be upgraded with search term matching and text customization switched on by default. Campaigns using the campaign-level broad match setting will receive search term matching by default, without text customization or final URL expansion.
That distinction matters. Automatically generated copy is visible and easy to debate. Search term expansion works earlier in the chain, before the ad appears in a report as a tidy row with a familiar keyword attached to it. It changes what the campaign can enter.
The upgrade also creates a strange asymmetry. A team may have opted into automated headline and description writing because it wanted creative variations. From September, that same account can inherit automated query expansion too. The original consent and the new behavior are related inside Google's product architecture, but they aren't the same business decision.
That is how platform defaults become strategy. One setting becomes evidence that the platform can safely make the next setting for you.
The query is the real audience
Broad match has always been sold as a way to find demand marketers didn't think to name. That can work. It can also blur the line between discovering an audience and redefining the audience.
A search campaign doesn't just answer the question, “Which keyword should trigger this ad?” It also answers, “What does this brand believe is close enough to the customer's intent?” Those are strategic questions, even when they appear beside a toggle.
Google's documentation describes a prioritization system where exact matches, phrase and broad matches, AI-based relevance, and Ad Rank influence which keyword is eligible for a query. That may be technically coherent. It doesn't remove the business problem. A campaign can be perfectly rational inside the auction and still attract searches that don't fit the offer, margin, geography, or brand promise.
The first audit should therefore happen at the query level, not the keyword level. Pull the last 90 days of search terms and sort them into four groups: exact commercial intent, useful discovery, expensive distraction, and brand risk.
The point isn't to build a giant negative keyword list and call the work finished. The point is to see what the system believes is adjacent to the business. The gaps tell you more than the hits.
This is closely related to the AI search visibility problem. Brands are learning that being present in an answer isn't the same as being represented accurately. Paid search now has a similar issue. Matching a query isn't the same as earning the right to influence that query.
Measurement gets harder before it gets better
The September upgrade arrives while many teams are still trying to understand what counts as an incremental conversion. More automated matching will produce more activity, not necessarily more clarity.
A platform can report that AI Max found additional conversions. The next question is whether those conversions were new demand, demand that would have arrived through another channel, or demand that was already close to purchase and simply got an ad attached to it. The interface may show a result. It won't settle the counterfactual for you.
This is where the attribution drift problem becomes operational. If a system uses its own reported conversions to improve bidding, and the same system determines which new queries and assets receive exposure, the feedback loop can reward the platform's interpretation of success without proving that the interpretation is right.
The danger is not a dramatic failure. It's a small shift repeated thousands of times. A slightly broader query. A slightly more generous conversion boundary. A little more spend on traffic that looks efficient in the platform view. By the time the blended result looks strange, the original change is buried in the account history.
Google also warns that AI Max will not perform well when a campaign is budget limited. That is not a minor footnote. Search term matching can widen the eligible query pool without widening the money available to fund it. A campaign may have more opportunities and less control over which opportunities consume the budget.
Tracking is the boring failure mode
The most expensive problems are often the ones that don't look like failures in the interface.
Final URL expansion can route users to dynamic landing pages instead of the fixed destinations a team chose. Tracking templates built around static URLs may behave differently. If the landing page, campaign, and analytics system disagree about where the user went, the account can report a clean conversion while the team loses confidence in the path that produced it.
That is why the September audit needs more than a list of enabled features. Test every important campaign from the ad click through the landing page, analytics event, CRM record, and revenue report. Check mobile behavior. Check redirects. Check UTM values. Check whether a generated asset sends traffic to a page that was never approved for that message.
None of this is glamorous. It is also where automated systems tend to expose the difference between a setting that works in a demo and a system that can be trusted with real demand.
What to do before September
Start with an inventory of affected campaigns. Find every campaign using the campaign-level broad match setting or automatically created assets. Record the current match behavior, pinned assets, landing pages, tracking templates, budget status, and conversion action.
Then create a before-and-after measurement view. Save the current search term mix, impression share, cost per conversion, conversion value, new customer rate, and downstream revenue. Don't rely on the platform's default comparison window. You need a baseline that survives the setting change.
Next, choose where you want control. Regulated categories, high-consideration products, narrow geographic services, and campaigns with strict landing page claims deserve tighter boundaries. If a campaign depends on pinned compliance language, confirm what remains pinned after the upgrade. The PPC Land report notes that broad match campaigns upgraded with search term matching alone keep their pinned assets, while the automatically created asset path brings a different set of defaults.
Finally, set an owner for the first 30 days. Someone should review query quality, asset changes, landing page destinations, and conversion quality every week. Not because the system is automatically bad. Because a change this broad deserves an adult in the room.
Google's own 2026 Marketing Live announcements make the direction clear. More parts of the ad system are being connected so the platform can make more decisions with less manual input. That may improve performance. It also makes old account structures harder to inspect.
The right response isn't to reject every automated feature. It is to stop confusing less visible work with less important work. Search strategy is still happening. The only question is whether your team is making it, or inheriting it.
The setting is not the strategy
Google AI Max may improve paid search for many accounts. It may also expose how few teams have written down the boundaries around their campaigns.
September 1 is a product deadline, but the deeper deadline is managerial. If the account cannot explain which searches it wants, which searches it refuses, what counts as a valuable conversion, and who can reverse a bad decision, the platform will fill in the blanks.
The campaign will still look optimized. That doesn't mean it will still belong to you.