What you'll learn in this article
- Why AI Max is not a rebrand of Performance Max and not a new campaign type at all
- What is AI Max in Google Ads in practice: a feature layer you switch on inside an existing Search campaign
- What its query expansion actually does, search term matching, keywordless recall, and the new "AI Max" match type in your reports
- The specific reasons I evaluate it with caution instead of enabling it account-wide by default
- The controlled way I turn it on, ad group by ad group, and the signals I watch before scaling
Almost every question I get about what is AI Max arrives already tangled with an assumption: that it is Performance Max with a new name, or that it is a shiny new campaign type I have to create from scratch. Both are wrong, and getting this straight is the whole point of the article. In the accounts I manage, the practitioners who struggle with AI Max are almost always the ones who treated it as something it isn't. So before touching a single setting, the honest answer to what is AI Max in Google Ads is narrower and more useful than the marketing suggests: it is an optional layer of AI-driven targeting and creative features that sits inside a Search campaign you already run. Understanding that boundary changes how you test it, and whether it burns budget or earns it.
AI Max is not a rebrand of Performance Max
Let me kill the two most common misreadings first, because they lead to real money lost. Performance Max is a separate campaign type that serves across Search, Display, YouTube, Gmail, Discover and Maps, and it deliberately hides most query-level detail from you. AI Max is the opposite in structure: it lives inside Search only, and it keeps your ad groups, your keywords, your negative lists and your search terms report intact. Google has admitted the two share a lot of underlying technology, but "shares technology" is not "is the same product". One replaces your campaign; the other layers onto it.
The second misread is treating it as a new campaign type. You do not open the campaign creation flow and pick "AI Max". You take a Search campaign, open its settings, and switch the feature on. That distinction matters for how you plan an account: if you are mapping out the broader AI Max campaign setting in Google Ads, you plan it as an enhancement to Search, not as a parallel channel competing for its own budget line. When both an AI Max Search campaign and a Performance Max campaign in the same account are eligible for a query, the candidate with the higher Ad Rank serves, the same logic that already governs broad match and PMax.
What AI Max really is: a feature layer inside Search
Stripped of the branding, AI Max bundles two engines plus a set of controls. The first engine is search term matching: it expands your reach beyond the keywords you entered, using broad match and keywordless technology to find queries Google predicts will convert. The second is asset optimization, which combines text customization (generative ad copy built from your existing ads, landing pages and assets) with final URL expansion, sending a click to the most relevant page on your domain rather than the URL you set. Turning AI Max on activates both, and you can toggle each off individually.
Around those two engines sit the controls that make AI Max defensible on a managed account: brand inclusions and exclusions, location of interest targeting at the ad group level, and URL inclusions and exclusions so you decide which landing pages the expansion may and may not use. Google's own documentation on how AI Max works lays out these features and confirms one detail every practitioner needs to internalise: final URL expansion requires text customization to be on, and when it is, pinning of your responsive search ad assets is no longer respected. If asset pinning is non-negotiable for compliance or brand reasons, that alone changes whether you can enable the full stack.
What the query expansion actually brings in
This is the part that answers what is AI Max in Google Ads at the level that shows up in your data. Search term matching does not just widen your broad match, it adds keywordless recall, meaning your ad can match a query with no keyword in your account that corresponds to it, driven instead by your landing pages, assets and campaign context. In the search terms report, these incremental queries are tagged with a dedicated "AI Max" match type, and a new source column tells you whether a match came from broad match expansion or from keywordless matching. That transparency is the feature I value most, because it is what makes the expansion auditable instead of a black box.
Practically, this behaves like a more aggressive relative of broad match: more reach, more irrelevant queries to prune, and a heavier dependence on Smart Bidding to sort intent. It is also the exact point where people conflate it with PMax, so if that comparison is what you're after, I break it down separately in AI Max vs Performance Max. AI Max leans on the same principle broad match does, that negative keywords stay your primary steering tool. Those negatives are still respected with AI Max on, but the volume of new, unfamiliar queries the expansion surfaces means your existing list is essentially never enough on day one.
Why I evaluate it with caution, not blind adoption
Google's headline number is a typical 14% lift in conversions or conversion value at a similar CPA or ROAS, rising to around 27% for campaigns still heavily reliant on exact and phrase match. I take those figures seriously, but I read the footnotes. The 14% benchmark explicitly excludes retail advertisers, which is a large, conspicuous carve-out, and independent testing across the industry has reported wildly inconsistent outcomes: many accounts see neutral or negative results, and at least one agency test found the overwhelming majority of AI Max impressions producing zero conversions across tens of thousands of terms.
None of that means AI Max is bad. It means the outcome depends heavily on vertical, account maturity, budget, landing page quality and how disciplined you are with exclusions, exactly the variables Google's average conveniently smooths over. On a well-optimised account, squeezing another 14% of volume without an efficiency hit is a strong claim that deserves testing, not faith. The other caution is mechanical: search term matching does not work under Manual CPC, because the system feeds on the signals a conversion-based bidding strategy provides, a dependency I cover in more depth in the walkthrough of how AI Max works for Search campaigns. If your bidding is not conversion-driven, the core of AI Max simply won't run.
How I actually test AI Max on live accounts
My rollout is deliberately narrow. I enable it first on non-brand, generic ad groups where incremental reach is genuinely useful, and I keep it off brand-defense ad groups, where the last thing I want is the algorithm reinterpreting a high-intent brand query. Because search term matching is an ad group level control once AI Max is on at the campaign, I can be surgical: expansion where I want exploration, keywords-only where I want protection.
Then I watch the data harder than usual. In the first few weeks I filter the search terms report to the "AI Max" match type every three to four days and add negatives immediately, without waiting for statistical significance, because the expansion surfaces waste faster than a normal campaign. I judge the test on incremental conversions from net-new queries, not on the aggregate campaign number, which can flatter a mediocre expansion. And I confirm the budget is adequate: starving it produces the worst of both worlds, wide matching with too little data for the bidding to learn. Treat it as a controlled experiment layered onto a healthy Search campaign, and AI Max earns its place or exposes itself quickly. Adopt it account-wide out of the box, and you are just paying to find out the same thing the slow, expensive way.
