Google’s AI Max Can Break Your Attribution

Google’s AI Max for Search can send a click to a different page on your site than the final URL you chose, so tracking built around that one URL can break, and the first sign may be leads reaching your CRM with no source. If you use a tracking template, Google’s guidance is to place the {lpurl} tag correctly and test an expanded landing page before you turn AI Max on, and URL exclusions keep expansion away from pages that should never receive ad traffic. We check this before launch because in our XRHealth case study, a tracking gap left about half of all leads without a known source, and paid campaigns looked weaker than they were until attribution coverage reached 99.9%.

The fix took two to three weeks, and our post on taking attribution from 50% to 99.9% explains how we found and closed the gap. This post covers the AI Max version of the same risk.

What Changed With AI Max, and When

Google announced AI Max for Search campaigns on May 6, 2025, as a beta rolling out to all advertisers globally starting later that month. It bundles three features: search term matching, text customization, and final URL expansion. Final URL expansion requires text customization and is on by default when you opt a campaign in to AI Max. With both on, Google may replace your final URL with a more relevant page on your domain, one that Google says will be query relevant and themed to the ad group, and generate headlines and descriptions to match it.

The second date is the one to act on. On April 15, 2026, Google said AI Max was moving out of beta and that, starting in September 2026, remaining eligible Search campaigns still using Dynamic Search Ads, automatically created assets, or the campaign-level broad match setting would be upgraded to AI Max automatically. Google said it expected those upgrades to finish by the end of September. On June 11, 2026, Google pushed the Dynamic Search Ads auto-upgrade back to February 2027, while campaigns using automatically created assets or the campaign-level broad match setting were still scheduled to upgrade starting in September 2026. Under the April plan, former Dynamic Search Ads campaigns would move over with all three features on, including final URL expansion, and with their legacy URL controls preserved. Campaigns that used automatically created assets would get search term matching and text customization by default, and campaign-level broad match campaigns would get search term matching by default.

If your account used automatically created assets or the campaign-level broad match setting, open the AI Max section of each Search campaign’s settings and confirm what is switched on, even if nobody on your team opted in. If it still runs Dynamic Search Ads, run the checklist below before you upgrade those campaigns yourself, and in any case before February 2027, when Google says remaining Dynamic Search Ads campaigns will be upgraded automatically. Legacy URL rule types such as “page title contains” and “page content contains” become read-only after the upgrade. You can remove them but not add or edit them, so add new exclusions as URLs or with the rule types AI Max supports.

Does AI Max Break UTM Tracking?

It can. Google does not say that AI Max strips UTM parameters, but it also does not say whether UTM tags typed into a final URL carry over when final URL expansion picks a different page, so test that before you rely on it. What Google’s help page does say is that, if tracking templates are not configured in a compatible way, dynamic landing pages from final URL expansion can sometimes break and return a 404 error. It names two setups that cause trouble:

  • Static tracking URLs. If your tracking template points to a fixed address with no {lpurl} tag, visitors can land on that fixed address instead of the page AI Max chose.
  • Nonstandard {lpurl} placement. If {lpurl} is only part of a parameter value, dynamic landing pages may not work as intended, and visitors can hit 404 errors.

The quieter risk sits downstream. A common CRM setup reads UTM parameters from the landing page URL into hidden form fields, sometimes only on the pages a campaign was built for. If AI Max sends a click to a page without that form or script, or the UTM tags you typed into one final URL do not appear on the expanded page, the lead reaches the CRM with no source. With auto-tagging on, Google Ads adds a click identifier (GCLID) to the ad click’s URL, and Google Analytics 4 gives auto-tagging priority over manual UTM tags. That means GA4 can keep crediting Google Ads while your CRM loses the source, so check both systems.

What Should You Check Before AI Max Goes Live?

This is the three-step checklist we use. Run it before you opt a campaign in, and again for any campaign Google upgraded automatically.

1. Record Your Attribution Baseline

Before anything changes, write down two numbers and one list: the share of leads or customers in your CRM that carry a known source and campaign, the share of sessions GA4 reports as direct or unassigned, and the landing pages your paid traffic uses today. If Google already upgraded the campaign, pull the same figures for a period before September 2026, such as August, and compare them with the weeks since. Without that baseline, a drop after launch looks like a performance problem instead of a tracking problem, and the wrong campaigns get cut.

2. Make Tracking Work on Any Page AI Max Picks

  • Use a compatible {lpurl} pattern. Google recommends {lpurl}?, {lpurl}&, or {lpurl}# when tracking parameters follow the landing page, and {lpurl} on its own when there are none. Replace any tracking template that hard-codes a fixed URL.
  • Move UTM tags off the final URL. Google describes the final URL suffix as parameters attached to the end of your landing page URL, and you can set it at the account, campaign, ad group, ad, and keyword levels. That makes it a better place for UTM tags than one hard-coded final URL, though Google does not say how it behaves on expanded pages, so check it in the test below.
  • Add URL exclusions. In the AI Max section of the campaign settings, you can exclude specific URLs, or write rules, so final URL expansion stays away from pages such as blog posts, legal pages, or any page without your lead form and tracking scripts.
  • Test before launch. Google advises that if a campaign uses a tracking template, you confirm that a dynamic landing page from final URL expansion works before you turn AI Max on. Load a few likely pages with your full tracking string appended, submit a test lead, and check that the UTM tags and the GCLID reach the CRM record, including through any redirects.
  • Turn off final URL expansion if a test fails. In the campaign’s AI Max settings, under Asset optimization, clear the Final URL expansion checkbox until your tracking passes. You can do this without turning off text customization, search term matching, or AI Max itself.
  • Watch the pages AI Max chooses. After launch, the landing pages report has a “Selected by” column for AI Max landing pages, and the search terms report shows headlines and URLs. Run the same test on any page you did not plan for.

3. Build Tracking That Does Not Depend on One URL

Do not rely on a single UTM string surviving the whole journey. In our XRHealth case study, the fix had four layers:

  1. Parameter preservation. Campaign parameters were protected through redirects and page transitions, so source and medium survived as visitors moved between landing pages, scheduling tools, and sign-up forms.
  2. A persistent ID. One ID linked the first website session to the CRM lead and the final customer record.
  3. Redundant checkpoints. Tracking checks at the landing page, form submission, and CRM entry meant that a lead that lost its source at one point could be reconnected at another.
  4. CRM source mapping. Acquisition data was written straight onto the lead record instead of living only in an analytics dashboard.

The first layer matters most with final URL expansion, because a visitor may land on a page you did not plan for and reach your form from there. For Google Ads traffic, also store the GCLID on the lead. Google Ads needs auto-tagging to import conversion data from your CRM, and the stored click ID gives you a second way back to the ad when UTM tags are missing. Build all of this inside your consent setup: a click ID or persistent ID stored on a lead record is personal data under the GDPR, so collect and use it in line with each visitor’s consent choices.

Treat Attribution as Infrastructure

When final URL expansion is on, Google chooses some of your landing pages, so tracking that only works on the pages you picked is a standing risk. We treat attribution as infrastructure, not reporting: if the source does not follow the lead into the CRM, CAC and LTV by channel are guesses. If you want a second check before a rollout, this audit is part of our web and analytics work.

Sources (checked October 23, 2026): Google, “Unlock next-level performance with AI Max for Search campaigns” (May 6, 2025); Google, “Dynamic Search Ads are upgrading to AI Max” (April 15, 2026); Google, “We’re upgrading Dynamic Search Ads to AI Max” (updated June 11, 2026); Google Ads Developer Blog, “Dynamic Search Ads (DSA) Automigration Delayed to February 2027 and Campaign Creation Restored” (June 11, 2026); Google Ads Help, “How AI Max for Search campaigns works,” “Set up AI Max for Search campaigns,” “About Final URL expansion in Search,” “Add a Final URL suffix,” and “About auto-tagging”; Regulation (EU) 2016/679 (GDPR), Article 4(1) and Recital 30.

Updated October 23, 2026: we rewrote this post, re-checked its facts, and added sources.

Picture of Alex Piliavsky

Alex Piliavsky

Alex Piliavsky is a co-founder of Alchemy Avenue, an AI-driven growth and lifecycle marketing agency. He helps businesses build growth that actually lasts instead of chasing quick wins. With over a decade bouncing between countries and industries, Alex learned the hard way that real growth isn’t about luck! It’s about patience, systems, and boring stuff like clarity. He specializes in strategic digital marketing and automation, turning chaotic operations into predictable growth machines.
Picture of Alex Piliavsky

Alex Piliavsky

Alex Piliavsky is a co-founder of Alchemy Avenue, an AI-driven growth and lifecycle marketing agency. He helps businesses build growth that actually lasts instead of chasing quick wins. With over a decade bouncing between countries and industries, Alex learned the hard way that real growth isn’t about luck! It’s about patience, systems, and boring stuff like clarity. He specializes in strategic digital marketing and automation, turning chaotic operations into predictable growth machines.

Related Articles

Hands moving gold coins from several small jars into one larger jar

Cut Fast or Burn Budget: A UA Lesson

Scaling UA depends on cohort-level measurement and cutting losing channels fast. In our Eyecon case study, monthly spend grew from about $10,000 to roughly $600,000, and payback fell from over a year to 2–3 months. Here are the rules.

Read More
Man in a dark office marking a large printed spreadsheet with a red pen, a laptop open to a product roadmap

User Demand Isn’t a Business Case

At a caller-ID app, around 25% of surveyed users said they’d want VoIP calling. We didn’t build it: a flat “unlimited” price couldn’t cover per-minute costs. Here are five checks to run before you build.

Read More