AI Max and Performance Max: What Advertisers Actually Control in Google Ads

I’ve spent almost 15 years in paid search, and I’ve watched Google make the same move over and over: open up targeting, hand more decisions to the algorithm, and frame it as progress. AI Max for Search and Performance Max are the latest version of that pattern, and if you manage Google Ads for a medical device, healthcare, or other regulated brand, they’re worth understanding in-depth, not simply adopting because Google says it’s time.

This isn’t an argument against using them. It’s a look at what these campaign types are actually built to do, why the “AI-powered” framing oversimplifies things, and what advertisers can still control when levers keep disappearing.

What Performance Max Actually Does

Performance Max and AI Max are two different products attempting to solve two different problems (Google’s problems, not necessarily advertisers’ problems), so let’s unpack how these campaigns have been marketed vs. how they actually work.

Performance Max, at its core, is a way for Google to sell inventory across its full network, Search, Display, YouTube, Gmail, Maps, Discover, in a single package that obscures how any individual placement is performing. Some of that inventory is truly valuable: Gmail and Maps placements tend to perform well because they reach high-intent audiences in a specific context. But Performance Max also includes Google’s partner networks, a set of third-party sites that can be low quality and difficult to distinguish from spam. It also includes Youtube – a format arguably not built for last-click conversion at all. In the good old days, advertisers had the ability to buy the good placements and skip the bad ones. Now, Performance Max bundles them and reports on them as one number.

The business reason driving these updates is not a mystery. Search volume and market share are under pressure, so Google has an incentive to move advertiser budgets into inventory that’s harder to sell on its own merits. Grouping it all together and calling it “AI-optimized” forces advertisers to purchase the bad placements along with the good, diluting the precision of budgets and, often, leading to CAC increases.

The upside is that Google has responded to advertiser pushback by adding more visibility into where Performance Max ads do show up. You can now see the placement breakdown in a way you couldn’t a couple of years ago. What you can’t do is exert the same level of control you had when these were separate, individually buyable channels.

What AI Max for Search Actually Does

AI Max, in contrast, increases auction competitiveness in search results specifically. Functionally, it works a lot like broad match keyword targeting. It takes the keywords you give it as a suggestion, then expands to anyone Google’s system judges to have “relevant intent.” In practice, that means you can be targeting a specific product category and start showing up against searches with only a loose connection to it. We’ve seen this firsthand with a medical device client: a campaign built around a specific delivery device started generating impressions and clicks tied to consumer wearables, simply because Google decided the intent was adjacent enough.

The result of broader match is usually cheaper clicks and a wider audience, alongside less precision and a real risk of inflated cost per acquisition if the account isn’t being watched closely.

Why “AI Max” Doesn’t Mean AI Overviews

To be frank, because the product name implies the opposite: adopting AI Max does not, by itself, put your ads in Google’s AI Overviews.

Google’s Ads Liaison has now confirmed the mechanics twice. Ads can serve above or below an AI Overview, or inside it, but not both in the same auction, and only broad match keywords or keywordless targeting are eligible to serve inside one. Exact and phrase match keywords aren’t eligible at all. In December, Google clarified a second point: an exact match keyword no longer blocks its broad match twin from serving in an AI Overview, because the exact match version was never competing for that placement to begin with.

Follow that to its conclusion and you get an awkward result. Google’s own AI Max guidance steers advertisers toward tight, exact match keyword sets and lets search term matching handle the expansion. But exact match is ineligible for the placement most teams adopted AI Max to reach. The reliable path into AI Overviews is broad match in a standard search campaign, something most accounts were already running before AI Max existed.

If your team turned on AI Max specifically for AI Overview visibility, that’s worth revisiting. And if you’re trying to build a deliberate strategy for AI search placement, the honest answer today is that a clean lever doesn’t exist yet. Until Google clarifies a path forward, where dollars get spent between traditional search and AI Overviews is a black box.

Why Google Is Pushing Broader Targeting

None of this is new behavior from Google, just the latest packaging. Google has used AI-based bidding for roughly a decade. What’s changed is the pressure to move advertisers away from manual, tightly segmented targeting and toward broad match and automated bidding, where Google’s systems make more of the decisions. When targeting loosens across the board, more advertisers end up competing in the same auctions, a dynamic known as auction depth, and Google benefits from that regardless of how any individual campaign performs. More advertisers bidding in the same ad auctions drives up cost-per-click for everyone.

Google has also shown a pattern of sunsetting the campaign types that don’t fit this direction. Dynamic Search Ads are being phased out in favor of AI Max, despite being a meaningfully different product. It’s a reasonable bet that traditional search campaigns eventually see the same treatment. Planning for this now, and building the required measurement infrastructure to power it, is the most future-proof strategy.

Why This Hits Regulated Industries Harder

Performance Max and AI Max both depend on the quality of the conversion signal you feed them. For e-commerce, that signal is usually strong. A purchase is a purchase, and Google gets a clean, immediate outcome to optimize toward. In healthcare, medical device, and other privacy-first industries, that signal is almost never that clean. Lead volume isn’t the same as a qualified prospect, and a qualified prospect isn’t the same as a sale. Without real outcome data flowing back to the platform, the algorithm starts making assumptions, and those assumptions don’t always drive toward true business objectives.

Compliance requirements make this harder still. Cookie rejection rates are high on healthcare websites, which means a sizeable share of pixel-based tracking data never reaches the ad platform in the first place. Regulated advertisers are being asked to compete on data quality in an environment specifically designed to limit the data they can collect.

What Advertisers Can Still Control

The game is totally different than it was 2, 5, 10 years ago. Here’s what advertisers (particularly those in regulated, lead-gen-based industries) need to dial in to be successful in this new era:

• Feed the platform better data. Offline conversion data, tied back to real outcomes rather than lead volume alone, is the single highest-leverage thing an advertiser can do here. Every client we’ve helped get closer to real outcome data has seen a meaningful improvement in return on ad spend.
• Exclude the search partner network. This is one of the few placement controls still available in Performance Max, and it’s worth using if lead quality is a concern.
• Invest in creative differentiation. When targeting gets broader and less precise, the ad itself has to work harder to reach and convert the right person.
• Treat site speed and landing page experience as competitive advantages. Load time and post-click conversion rate directly affect cost per acquisition and auction performance, and they’re two of the most overlooked levers in a regulated industry that tends to think of “the ad” as the whole job.

The businesses that succeed in this environment are the ones that can turn messy, real-world data into something an ad platform can actually use, and that increasingly means having a clean data infrastructure on the business side to begin with. Asking an agency to fix targeting without first fixing the underlying tracking data is asking them to spend toward results they can’t see.

How to Test Without Risking the Budget

Adopting AI Max and Performance Max isn’t optional forever. These campaign types are the direction Google is heading, and learning how to use them well now is more useful than resisting the inevitable. That said, the advertisers who get burned are usually the ones who go all in on Google’s recommendation before proving the model works for their business.

A better approach is to allocate a defined test budget, small enough that a poor result doesn’t damage the broader account, and prove the mechanics work before pressing the gas pedal. Get the data pipeline right first, then scale.

The Real Shift: From Keyword Targeting to Data Quality

The line between marketing analytics and digital advertising has gotten thin enough that it’s fair to say they’re becoming the same discipline. Manually finding the right keywords and steering the right audience toward your ads, the skill set that defined paid search since its inception, matters less every year. What matters today is whether your business can identify a real outcome, tie it back to the click that drove it, and hand that signal to the platform in a form it can use.

That’s a harder, less visible skill than keyword research, and it’s not always an easy case to make to leadership. It requires investment in operations, IT, data infrastructure cleanup – all historically outside of marketing’s direct purview. But now, it’s the most important determinant of performance, in Google Ads and everywhere else the platforms are headed – and we’re not going back anytime soon.