Google has officially announced the sunset of Dynamic Search Ads, and the replacement they’re pointing you toward is AI Max for Search. If you’ve been running DSAs as your catch-all layer, as a gap-filler for long-tail traffic, or as the structural workhorse underneath a tightly managed keyword campaign, this is a change worth thinking through carefully before you do anything.
Why Dynamic Search Ads worked for so long
DSAs have been around since 2011. They worked because they were simple: point Google at your website, write your descriptions, let it match queries to relevant landing pages and generate headlines from your content. The mechanism was transparent enough that most practitioners understood what they were giving up (control over headlines and keyword targeting) in exchange for what they were getting (coverage they hadn’t mapped manually). For accounts with large inventory, complex service offerings, or constantly changing page content, that tradeoff made sense. You knew what the system was doing and roughly why.
AI Max is not a like-for-Like replacement
AI Max is not that. It is a more capable tool, and the performance claims are real in some contexts, but it operates differently in ways that matter. Where DSAs are matched based on your page content against a user’s typed query, AI Max determines relevance based on inferred intent. Google has been explicit about this: it’s matching based on what the system believes the user wanted, not what they typed. Brad Geddes documented the practical consequence of this when he found that AI Max treats all keywords as broad match, regardless of their actual match type, and attributes conversions to AI Max that would have been captured by your existing exact and phrase match keywords anyway. That means the incrementality numbers Google shows you need to be interrogated before you act on them.
Signal quality is the first thing to audit
The signal quality question is the one I’d start with. DSAs were predictable enough that most advertisers knew what they were measuring. AI Max introduces enough ambiguity into the search term report that you can end up in a situation where performance looks correct at the reporting level while the system has already updated its understanding of what a good outcome looks like, using data you wouldn’t have approved as a signal.
What to Check before migrating to AI Max
✅ If you’re migrating DSA campaigns to AI Max, audit your conversion setup before you flip anything. AI Max will go looking for what you’ve defined as a win. If your primary conversions are phone calls where the quality is questionable, it will optimize toward phone calls with questionable quality, faster and at greater scale than your DSA campaigns ever did.
✅ Exact match keywords in the same account still matter. They act as guardrails for your highest-intent queries and help establish what “good” looks like before you introduce broader matching behavior. Don’t strip them out in the name of consolidation.
✅ Review your Search Partners settings before enabling AI Max. This is not the moment to leave that checkbox alone. Research conducted by Mike Ryan has flagged that Search Partners sees a disproportionate share of sophisticated invalid clicks, and AI Max’s broader matching behavior means budget can end up there faster than you’d expect.
✅ Negative keyword lists should be built categorically, not reactively. The migration window is your opportunity to do this properly. Identify the categories of searches that will never represent your customer and exclude them before the system explores them.
✅ If you use DSA auto targets to exclude specific pages, map those exclusions into your AI Max text guidelines and URL rules before migrating. The exclusion logic doesn’t transfer automatically.
AI Max can work — but only with lean inputs
The honest assessment of AI Max is that it works best when your signals are clean, your conversion tracking is tight, and you have enough data volume for the system to learn accurately. An independent study conducted by the folks at Smarter Ecommerce, testing across more than 250 retail campaigns, found it delivered conversions at roughly 35% lower ROAS than traditional match types. That doesn’t mean don’t use it; it means the inputs matter more than the tool.
The real lesson: AI will follow your instructions
DSAs gave you relatively predictable coverage with a known set of tradeoffs. AI Max gives you broader reach with less visibility into how it’s making decisions. That’s not a reason to resist the migration; it’s a reason to be deliberate about what you feed it and what you tell it to ignore.
The system will follow your instructions. Make sure your instructions are right.
