AI Max for Search: Google's Query-Matching Overhaul

AI Max for Search is an opt-in feature suite inside ordinary Search campaigns — not a new campaign type. Switch it on and Google's AI gets three new permissions: to match your ads to queries beyond your keyword list, to generate ad text variations from your landing pages and existing ads, and to send clicks to pages on your site beyond your stated final URLs. Google introduced it in 2025 as the successor to a decade of separate expansion features, and it amounts to the largest control trade Search has offered since broad match.

The trade is real in both directions: reach into query territory your keywords never covered, in exchange for a campaign whose matching, text and landing pages are all partly machine-chosen.

How AI Max works

Feature	What it does	Control that remains
Search term matching	Matches ads to queries your keywords do not cover, using intent understanding rather than keyword text	Negative keywords; brand inclusions and exclusions
Text customization	Generates headlines and descriptions from your landing pages, ads and assets	Feature toggle; reporting shows generated text
Final URL expansion	Routes clicks to the page on your site it judges best for the query	Feature toggle; URL exclusions

Your keywords do not disappear. Google has said queries matching your keywords keep routing through them — the expansion applies where no keyword covers the intent, which is why AI Max behaves less like a match-type change and more like a second, keywordless campaign sharing your budget. The match types you know still describe the covered territory; AI Max annexes the uncovered part.

The Dynamic Search Ads connection is worth naming: final URL expansion does what DSA did, inside a standard campaign, and Google has positioned AI Max as DSA's successor, with upgrade notices appearing in accounts. Timelines have shifted during the rollout — verify what your account says rather than trusting third-party dates; the DSA-to-AI-Max migration walkthrough covers the moving parts.

Accountability lives in reporting: the search terms report labels AI Max matches so you can see exactly what the expansion bought, and asset reporting shows which generated headlines and landing pages served. The feature is only as reviewable as your habit of reading both.

Reading your own account before you opt in

The opt-in is an analysis before it is a toggle, and the analysis is readable from your own account: how much query territory do your keywords actually cover, what did expansion features historically do to your cost per conversion, how exposed is your brand to loose matching. An agent with account access can assemble that picture on request — the full decision framework is in the AI Max opt-in decision.

After opt-in, the job becomes supervision, and it is weekly, not annual: pull the AI Max-matched terms, compare their conversion economics against keyword-matched traffic, and turn the losers into negatives. Negative keywords are the working control lever for the matching layer, and an agent can run the whole loop — read, propose, and add negatives on approval. Generated text and swapped landing pages get the same treatment: read what actually served, flag anything off-brand, exclude URLs that should never receive paid clicks.

Opting in without that supervision loop is not using AI Max — it is being used by it.