Google's AI Max for Search launched globally in September 2025 with bold claims: 14% more conversions at similar costs. Four months in, the picture is more nuanced, especially for B2B advertisers.
Here's what we're actually seeing, why results vary so much, and the strategic reason you might enable it regardless.
What AI Max Actually Does
AI Max is a feature toggle within your existing Search campaigns. When enabled, three things happen:
1. Search Term Matching Expands
Google uses "keywordless" technology to match your ads to queries beyond your keyword list. Your exact match keywords still take priority, but AI Max adds incremental queries it believes are relevant. You can see in the search terms report whether a query was matched via AI Max or your traditional keywords.
2. Text Gets Generated
Headlines and descriptions are dynamically created based on your landing page, existing ads, and the query. This is the renamed version of "automatically created assets."
3. URLs May Change
Final URL expansion (enabled by default) lets Google redirect clicks to different pages on your site if it predicts better performance.
What Google Claims vs. What We're Seeing
Google's Numbers
The official marketing is compelling:
These numbers come from Google's testing and select case studies, likely with significant Google support and ideal conditions.
Independent Advertiser Data
The picture is mixed. Agency testing at Mediassociates found CPAs running "two to three times higher" than standard search campaigns. One advertiser's analysis showed 90% higher cost per conversion vs. phrase match. Analysis of 250+ campaigns found AI Max delivering 35% lower ROAS than traditional match types.
Our Experience
We're currently running AI Max on a B2B account. The honest assessment: no incremental lift whatsoever. The AI Max-attributed search terms in our reports aren't delivering conversions that move the needle. This matches what we're hearing from other B2B practitioners: the automation works better for high-volume consumer accounts than for B2B with longer sales cycles and lower conversion volume.
The Strategic Case: AI Mode and AI Overview Eligibility
Here's where it gets interesting. Even with flat performance, there's a strategic reason to enable AI Max.
Google now shows ads in AI Overviews and is testing ads in AI Mode (the Gemini-powered search experience). To be eligible for these placements, you need AI-powered targeting: broad match, AI Max, Performance Max, or Dynamic Search Ads.
Without AI Max or broad match, you're locked out of AI Overview and AI Mode ad placements entirely.
This matters because AI Mode is where search is heading. Ads in AI Mode are being pulled from Search campaigns with Broad Match, including AI Max. If you're running pure exact/phrase match, you won't appear in these formats as they scale.
So the calculation becomes: Are you willing to accept potentially flat (or worse) performance today to maintain eligibility for where search is going?
How to Evaluate AI Max in Your Account
You can't run a true A/B experiment: AI Max is either on or off for the campaign. But you can measure its impact:
Use the Search Terms Report
The search terms report now shows whether a query was matched via AI Max or traditional matching. Filter for AI Max-attributed terms and evaluate:
Disable What You Don't Want
AI Max has multiple components you can control:
Start with search term matching alone. If that works, layer in other features.
Set Clear Success Criteria
Before evaluating, define what success looks like:
Review after 30+ days with sufficient data.
When AI Max Makes Sense
When to Be Cautious
The Bottom Line
AI Max isn't the leap forward Google's marketing suggests, but it's not worthless either. For most B2B accounts, expect flat to marginal results on direct performance.
The real question is strategic: Do you want eligibility for AI Mode and AI Overview placements as Google shifts search toward AI-powered experiences? If yes, enabling AI Max (even with flat performance) is the price of admission.
Test it. Measure the AI Max-attributed terms. Disable the features that don't work. But don't ignore it entirely. You may be opting out of where search is going.