Google just announced that AI Max for Search will automatically expand your URLs. That sounds like a convenience feature. It isn’t. It strips your UTM parameters in the process — and if you’re running any kind of CRM-connected attribution pipeline, that’s not a minor inconvenience. That’s a blindfold.
This news reminded me of a story my partner Ohad told me from his time at a prominent digital health company. When he joined, attribution was sitting at 50%. Half of all users were showing up as “direct” traffic in Google Analytics and Mixpanel. With all due respect to the brand, there was no way it was generating that much organic pull. Something was broken. The data just didn’t know it yet.
What Broken Attribution Actually Costs You
The consequences weren’t abstract. The company was actively spending on Facebook and Google Ads. But because only half the data was visible, the campaigns looked like they were underperforming. CAC calculations were wrong. LTV per channel was a guess. Every optimization decision was being made on corrupted inputs.
You can’t fix what you can’t see. And when you can’t see it, you don’t cut the right things — you cut the wrong ones. That’s how good campaigns get killed and bad ones get funded.
Ohad started pulling the thread. The biggest culprit was Facebook’s link shimming — a process where users get redirected through an intermediary URL before landing on the site, which quietly strips UTM parameters along the way. He found a workaround that preserved them. He also built unique ID checkpoints throughout the funnel, so every user could be traced back to their origin point even if attribution broke down somewhere in the middle.
In two to three weeks, attribution went from 50% to over 99%. Leadership finally had a clear picture of what was actually working. Real CAC. Real LTV. Real decisions.
Google’s AI Max Is the Same Problem at Scale
What Google is doing with AI Max isn’t malicious. It’s just careless — at least from a marketer’s perspective. The system automatically rewrites destination URLs to improve ad relevance. The byproduct is that your carefully constructed UTM strings get overwritten or dropped entirely.
For teams running basic GA4 dashboards, this is annoying. For teams with CRM-integrated attribution models — where every click needs to carry a source, medium, and campaign tag all the way through to a closed deal or a subscription — this is a genuine crisis.
The “direct” traffic bucket is about to get a lot bigger for a lot of companies. And most of them won’t notice until the CAC numbers stop making sense.
What You Can Do Right Now
First, audit your current UTM coverage before the transition hits. Know your baseline. If you’re already at 70% attribution, you need to know that before it drops to 40%.
Second, look at your URL expansion settings inside Google Ads. AI Max gives you some controls — use them. Don’t let the default setting make decisions for you.
Third, build redundancy into your tracking. Don’t rely on a single UTM string surviving the full journey. Use unique ID parameters, server-side tracking where possible, and CRM-level source tagging that doesn’t depend entirely on what Google passes through.
Ohad pushed attribution from 50% to 99% by treating it as a systems problem, not a reporting problem. That’s the right frame. UTM parameters aren’t just labels — they’re the connective tissue between your ad spend and your revenue data. When they break, everything downstream breaks with them.
Google’s AI Max is coming whether you’re ready or not. The question is whether your attribution infrastructure can survive it.
