Google is retiring Dynamic Search Ads, and if you still run DSA campaigns the change is now visible in your account. In-account notices encouraging a voluntary upgrade to AI Max for Search started appearing in September 2026. The important detail most of the coverage blurs is the timeline: campaign-level Broad Match and Automatically Created Assets campaigns are the ones that auto-migrated in September 2026, while DSA campaigns specifically are not force-upgraded until February 1–28, 2027, at which point the ability to create new dynamic ad groups is removed for good. So you have a real prep window, not an emergency — and the worst thing you can do with it is nothing.
This post is not a generic “how to set up AI Max” walkthrough. It is about the one thing the migration changes that a practitioner has to get ahead of: what happens to your search terms and your negative coverage when a page-matched DSA campaign becomes a keyword-expanded AI Max campaign. Your negatives carry over, but the matching they were sized to contain gets wider. Handle that before February 2027 and the switch is a non-event. Ignore it and you inherit a search terms report filling with queries your old DSA list never had to block.
What actually changes in the matching
DSA never used keywords. It matched queries to your landing pages and site content, deciding which searches were relevant by reading your pages rather than by a keyword list you maintained. That is why DSA campaigns always leaned on negatives so heavily — the negative list was the only steering wheel you had, because you could not add keywords to shape what matched in. AI Max keeps the page-based matching but layers keyword-style query expansion on top, using AI to match searches it judges related to your content and to any keywords in the campaign. The practical effect is a wider funnel: queries DSA would never have connected to your pages can now arrive through expansion.
This is the same loosening that AI Max brings to ordinary Search campaigns, where AI Max undermines the match-type control advertisers rely on, and it is documented behaviour rather than a theory — expansion is the point of the format. For a DSA advertiser the shift is starker than for a keyword campaign, because you are going from no keyword matching at all to AI-driven matching, so there is no existing keyword-level containment to fall back on. The only containment you have is your negative list, and it was built for the narrower page-based world. That gap is the whole risk of the migration.
Audit the DSA search terms report before you touch anything
Start where every negative decision starts: the search terms report. Pull the full report for each DSA campaign over the longest window you can, sorted by cost, and read it as a record of what page-based matching has been quietly paying for. DSA campaigns accumulate a distinctive kind of waste — queries that matched a page technically about the right topic but at the wrong intent, or that landed on a thin or outdated page you forgot was still crawlable. Those rows are gold, because the themes DSA wasted money on are precisely the themes AI Max expansion will amplify once it turns on.
Work the report the way you would for any hidden wasted spend audit: filter to zero-conversion queries with real cost, group them into recurring patterns rather than picking off individual terms, and note the themes that repeat. You are not just cleaning up the current campaign; you are building the evidence base for the negative list that will protect the AI Max version. Every recurring waste theme in the DSA report is a negative you want in place before expansion starts producing more of it, so treat this export as the single most valuable artefact you carry across the migration.
Turn that report into a shared negative list
Do not add the negatives you find campaign-by-campaign. Build a shared negative list from the recurring waste themes and apply it, because the migration is the moment to consolidate rather than fragment. A shared list gives you one place to maintain the exclusions that should apply across every campaign that inherits your DSA traffic, and it means the containment travels with you into AI Max in a single, auditable object instead of being scattered across campaign settings you will forget you changed. The mechanics of choosing account-level, campaign-level and shared scope are covered in how to structure negative keyword lists, and they apply directly here.
Building the list before migration rather than after is the difference between inheriting your knowledge and restarting the waste clock. This is the same logic as maintaining proactive negative keyword lists before the spend happens: the cheapest waste to prevent is the waste you already know is coming. You have months of DSA search-term history telling you exactly which queries do not convert. Encoding that into a shared list before AI Max expansion goes live means the wider matching launches into a account that already knows what to reject, instead of one discovering it the expensive way over the following weeks.
Migrate a small campaign first and watch it daily
Rather than waiting for the automatic February 2027 switch to hit every DSA campaign at once, upgrade one low-spend DSA campaign manually and watch its search terms report daily for a week. This is a controlled test that answers the only question that matters for your specific account: how much does expansion widen the query set here? Some accounts barely move because their pages are tightly themed; others see a noticeable influx of loosely-related queries. You cannot know which you are without watching one campaign make the transition under observation, and a small campaign limits the cost of finding out.
Use that week to calibrate. Every new query type that expansion introduces is a signal about what your larger campaigns will see when they migrate, and each one you decide is waste goes straight into the shared list you already built — strengthening the containment before the high-spend campaigns switch. This staged approach also protects you from the trap of over-negating and throttling good traffic, because you are adding negatives against observed waste rather than guessing defensively. By the time your important campaigns migrate, the list is tuned to your real traffic and you know what “normal” looks like after expansion.
Expect the report itself to read differently
After migration the search terms report is still your control surface, but the terms in it read differently. AI-mediated matching means a reported query can reflect Google's interpretation of intent rather than the literal words a user typed, and low-volume queries stay redacted, so the report is both wider and slightly fuzzier than the clean DSA query log you are used to. That is not a reason to trust it less; it is a reason to review it more often in the weeks around the switch, when the newly-expanded matching is settling and producing query types the campaign has never seen.
Set a tighter review cadence for the migration period specifically — weekly at minimum, more often for high-spend campaigns — then relax it once the query set stabilises. The behaviour of AI-driven assets under expansion is covered in more depth in how AI Max automated assets inherit query expansion, which is worth reading alongside this because assets and matching interact: an expanded query set feeds automatically-generated assets you did not write. The report is where both surface, which is exactly why it stays the first thing you open every review through the migration and after it.
The pre-migration checklist
Condensed to a sequence you can run this week: export the full DSA search terms report per campaign; group recurring zero-conversion waste into themes; build a shared negative list from those themes and apply it; manually migrate one low-spend campaign and watch it daily for a week; feed every new waste query the test surfaces back into the shared list; then migrate the rest with the list already protecting them. None of this depends on the February 2027 deadline arriving — it is all doable now, and doing it now is the entire advantage of the warning window Google has given you.
The broader migration involves more than search terms — bid strategy, asset review and brand controls all deserve attention, and the full sequence is laid out in the AI Max migration search terms checklist and the post-migration negative keyword audit. But the negative list is the piece that determines whether the report fills with paid-for junk on day one or stays clean. Everything else can be tuned after the switch; the containment has to exist before it. Build it while you still have a DSA report to build it from.