Google announced AI Max for Search campaigns back in May 2025 at Google Marketing Live, rolling it out globally in beta shortly after. I’d been reading about it since launch, keeping an eye on it but I never had a reason to actually run it myself. Honestly, I wouldn’t enable AI Max on a fresh account without a solid data foundation in place first. It’s not the kind of feature you just switch on and hope for the best.
That changed in April 2026, nearly a year after it launched, when a client handed me their Google Ads account to manage. The campaign was already active and AI Max was already switched on.
So this is my first real, hands-on experience with it in a live environment. Not by choice, but by circumstance.
Here’s what I found, what I decided to do, and what I’m planning next.
A Bit of Context: What Is AI Max for Search?
For anyone who hasn’t come across it yet, AI Max is Google’s one-click feature suite that layers AI-powered targeting and creative enhancements on top of your existing Search campaigns. It combines:
- Search term matching — expands your keywords using broad match and keywordless technology to find queries you’d otherwise miss
- Text customisation — automatically generates new headlines and descriptions based on your landing page and keywords (formerly called “automatically created assets”)
- Final URL expansion — sends users to the most relevant page on your site rather than a fixed destination
- Granular controls — including locations of interest and brand controls at the ad group level
Google’s own data suggests advertisers typically see 14% more conversions or conversion value at a similar CPA/ROAS after enabling it — and for campaigns still heavily reliant on exact and phrase match, that number reportedly jumps to 27%.
So on paper, it looks compelling. But there’s more to the story.
How I Actually Started Using It — I Didn’t Turn It On Myself
As I mentioned, I didn’t enable AI Max on this account. When the client passed the account over to me in April 2026, it was already running with AI Max switched on an active campaign, live spend, and AI-driven targeting already in motion.
This is actually a fairly common scenario for anyone taking over a managed account. You inherit settings you didn’t configure, and your first job is to understand what’s running and whether it makes sense for where the account currently is — not just leave things running because someone before you set it up that way.
What I Observed
Diving into the search terms report, AI Max had clearly been doing its job of expanding reach. And to give credit where it’s due it was reaching a lot of people. The impression and click volumes were up, and the campaign was surfacing queries well beyond what the keyword list alone would have captured.
But here’s the problem: a significant chunk of that reach was going to the wrong audience entirely.
AI Max was spending budget on queries and users that had little to no relevance to the product. The targeting had casted the net wide wider than it should have for this particular niche and the account was essentially paying for traffic that was never going to convert. It’s one of those situations where the numbers look active on the surface, but when you dig into the quality of that activity, it tells a very different story.
The text customisation feature had also generated asset variations, and some of these were decent. But the quality of AI-generated assets is heavily tied to the quality of the landing pages and the existing ad copy and there was room for improvement on both fronts.
The core issue underpinning all of this: the account didn’t yet have enough conversion data to give the AI a strong signal to work with. AI Max like Smart Bidding and Performance Max is a multiplier. It gets smarter as it accumulates data about what a good conversion looks like for your specific account. Without that foundation, the system has no reliable way to distinguish a valuable audience from an irrelevant one. So it reaches broadly, spends freely, and waits to learn and that learning comes at a cost.
Why I Decided to Pause It
Given that the account was still in an early data-building phase, I made the decision to pause AI Max for now.
This wasn’t a verdict on the feature itself. It was a timing call.
AI-powered tools in Google Ads tend to reward patience and preparation. The more conversion history the AI has — which queries led to sign-ups, which landing pages performed, which audiences converted — the better its targeting decisions become. Switching on a broad, AI-driven expansion tool before that foundation is in place risks inflating spend on poorly matched queries before the system has enough signal to self-correct.
For a B2B SaaS campaign in a niche vertical, where conversion volumes are naturally lower, this matters a lot more than it would for a high-volume e-commerce account.
What I’m Doing in the Meantime
The plan right now is to build the data foundation that AI Max needs to work well:
- Accumulating conversion volume — getting consistent conversion events (sign-ups, demo requests, or meaningful micro-conversions) to give the AI something solid to learn from
- Strengthening the keyword base — building out well-performing exact and phrase match data that AI Max can use as a reference point when it starts expanding
- Tightening up landing pages — since AI Max uses landing page content to generate assets and guide URL expansion, having well-optimised, content-rich pages is more important than ever
Once the account reaches a comfortable conversion threshold ideally 30–50 conversions per month consistently I plan to re-enable AI Max and give it a proper run with a strong baseline to learn from.
What This Experience Taught Me About AI Max
Even though my introduction to AI Max wasn’t by choice, it was a useful one. A few things stand out:
- Timing matters more than the feature. AI Max is not a magic switch. It’s a tool that amplifies what’s already working. If the data foundation isn’t there, the results will reflect that.
- Inheriting an account with AI-powered features already active is more common than you’d think — and the first thing to do is assess whether those features are appropriate for where the account currently stands, not just leave them running because someone before you turned them on.
- The controls are genuinely useful. Brand controls and locations of interest give you guardrails that were previously only achievable through careful keyword sculpting good to know for when I turn it back on.
- AI Max is still very new. Given it only launched in mid-2025, there’s still a lot to learn about how it performs across different verticals and account types. The case studies so far like L’Oréal’s 2X conversion rate improvement or MyConnect’s 16% lead increase are encouraging, but they’re from accounts that likely had strong data foundations in place.
What’s Next
This is very much a story in progress. Once I’ve built up the data the account needs, AI Max goes back on and I’ll document the results properly.
The potential is clearly there. I just want to give it the best possible chance of delivering on that potential, rather than activating it too early and writing it off unfairly.
Have you encountered AI Max in an account you’ve taken over, or run your own experiments with it? I’d be curious to hear how it’s gone drop a comment below or find me on LinkedIn.
