Google's AI Max passed its first birthday last week, and with it came a feature most teams have not reckoned with yet: Final URL expansion now supports mandatory text disclaimers. Translation — your ad can land on a URL Google chose, not the one you wrote, and that page now has to render a piece of compliance text it did not have an hour ago. The teams that ran clean message-match audits a quarter ago are working with assumptions that no longer hold. The audit needs to happen again, fast, and with a different scope.
Diagnosis: three things changed at once
The root cause is structural. Final URL expansion shifts the destination URL from a fixed campaign asset to an algorithmically chosen one. The ad and landing page used to be set up as a pair — same headline, same offer, same visual cues — and the alignment, if it existed, existed by design. Now the pair is broken. Your campaign points at a parent URL; Google's AI Max picks among children. The pages it picks may be older, slower, less aligned, or written for a different audience.
Then layer in mandatory disclaimers via Final URL expansion. The disclaimer is required to be rendered on the page. If your landing page does not have it, the system either injects it or fails the page. In either case, the page the user sees is not the page your designer signed off on.
Most teams miss this because the legacy Quality Score number does not appear in AI Max campaigns as a visible metric. The signals underneath — landing page experience, ad relevance, expected CTR — are still operating, still influencing how Google distributes your budget. But you cannot see the number that used to tell you the alignment was off. You only see the consequence: higher CPCs, lower share of voice, weaker conversion rate, and a vague feeling that the campaign is drifting.
This is the alignment drift problem in its sharpest current form: a system that picks its own destinations, on a platform that no longer shows you the score, with required text changes you did not write.
The AdAlign approach: three coordinated audits, not one
AdAlign treats this as three coordinated audits, not one. Upload an ad screenshot — your headline, your visual, your offer — and paste the campaign's final URL pattern. AdAlign then walks the children Google's AI Max may expand into and scores each one across three dimensions: visual match (does the landing page hero echo the ad's visual language?), content match (does the headline restate the ad's promise within the first viewport?), and tone match (is the page written in the same register — urgent, considered, technical — as the ad?).
Each child URL gets a score from 1 to 10. Below 7, AdAlign returns specific mismatches: "the ad promises a 14-day free trial; the landing page hides it under FAQ #4," or "the ad uses concrete numbers; the page leads with abstract benefits," or "the ad's visual is a product screenshot; the page leads with a stock photo of a team meeting." These are the fixes you can ship in a sprint.
For the disclaimer layer, AdAlign flags pages where the mandatory text either is not present (the system will inject it, often breaking your hero layout) or is present but breaks message match (a 90-word compliance block above the fold that pushes the offer below it). Agencies typically find three to five critical mismatches per client on a first audit, and at least one of those mismatches is now a disclaimer-and-layout conflict that did not exist three months ago.
The output is a prioritized fix list. Highest-impact fixes (above-the-fold headline rewrites, hero image swaps, disclaimer placement) sit at the top; cosmetic mismatches (button color, supporting copy) sit at the bottom. The whole audit runs in minutes, not days, which is what makes the 48-hour timeline real rather than aspirational.
Hour by hour: what the 48 hours actually look like
A useful sanity check: if you cannot describe the 48 hours concretely, the timeline is marketing copy, not a plan. Here is the breakdown that has held up across the agencies running this play.
Hours 0–4 sit on the audit itself. Export the expanded URL set from your top three AI Max campaigns. Run each ad-URL pair through AdAlign. You should finish with a scored list of every (ad × expanded URL) pair, ranked by congruence score, with the disclaimer-injection failures called out separately.
Hours 4–16 are the writing and design pass. Take the bottom decile — pages scoring below 6 — and fix the hero. In most cases this is a one-sentence headline rewrite plus a hero image swap. The pages between 6 and 7 are worth a second pass; pages above 7 are not worth touching this week. Do not boil the ocean.
Hours 16–32 are the disclaimer layout pass. For each page where the mandatory text either is not present or is breaking the hero, decide whether to embed the disclaimer in a fixed location (cleaner, slightly more work) or to redesign the hero to accommodate the injected version (faster, slightly less clean). Either is defensible. Mixing strategies across pages is not.
Hours 32–48 are the verification cycle: re-score every fixed page in AdAlign, push the rebuilt pages, and watch the next 24 hours of campaign data for share-of-impression recovery. The teams that skip this step end up redoing the work three weeks later when they cannot tell whether the fix worked.
Real-world application: the warning signs are quiet
The warning signs of AI Max message-match drift are quiet and specific. First: a campaign that performed well in its first month starts losing share of impression without any change to bids or budget. Google's AI Max is reading the landing pages and rebalancing. Second: CTR holds steady but conversion rate slumps. The ad keeps winning the click; the page stops delivering the promise. Third: Quality Score visibility (where you still get it, in search campaigns running alongside AI Max) drops on landing page experience while the other components stay intact.
If you run AI Max with Final URL expansion enabled — which Google has been auto-enrolling existing search campaigns into — you almost certainly have at least one campaign with drift right now. The question is how many, and where the highest-impact fixes sit.
The most expensive variant of this problem is when an agency's client portfolio runs AI Max across dozens of accounts. Manual checks at that scale are not done because they cannot be done. Drift accumulates across the portfolio, the agency takes the blame for performance erosion, and the actual cause is invisible. The Agency tier and agency workspace handle that portfolio-wide picture in one place; the related read on why ads get clicks but landing pages do not convert covers the diagnostic frame in more depth.
This is the moment the governance gap stops being theoretical. Creative is automated. Destination URL selection is automated. Required text injection is automated. The one thing nobody is automating is the check that all three still tell the same story. That is the gap — and the larger version of it, where the page itself starts to leave the post-click path, is what Universal Cart and the future of ad-to-page governance covers.
Run the audit
Run AdAlign's free audit on your top AI Max campaign. Three analyses, no card. You will see exactly which expanded URLs are drifting, where the disclaimer placement breaks your offer, and which fixes recover the most performance per hour of work.