On this page
Dynamic Search Ads Deprecated: What Is Happening
Key Takeaway: Dynamic Search Ads is being deprecated from September 2026. New DSA campaigns can no longer be created through the Google Ads interface, Google Ads Editor or the API, and existing DSA campaigns are being upgraded into AI Max for Search. This is a forced migration with a date, not an optional beta you can decline.
AI Max for Search exited beta on 15 April 2026 with a migration path that reaches every advertiser by September. The part that catches people out is the creation block: once it lands, you cannot spin up a new DSA campaign anywhere, including programmatically. Teams that build campaigns via the API are the most exposed, because a script that has run fine for years will simply start failing.
If you run Dynamic Search Ads today, you have a narrow window to migrate deliberately rather than have the migration happen to you. The difference between those two outcomes is mostly preparation, and most of the preparation is unglamorous.
Before we get into it, one framing point. This is not Google removing a feature out of spite. DSA's original job — find queries your keyword lists missed, using your own site content as the map — is exactly the job a modern model does better. The upgrade is genuinely more capable. It is also considerably harder to control, and that is where the work is.
What Actually Changes with AI Max?
DSA and AI Max share a premise: let the system find relevant queries you did not explicitly list. They differ in how wide they cast and how much you can constrain them.
Dimension | Dynamic Search Ads | AI Max for Search |
|---|---|---|
What it targets from | Your landing pages and site content, or page feeds you supply | Site content plus broader intent signals the model infers |
Query breadth | Narrow — bounded by what your pages actually say | Substantially wider. It casts a wider net by design |
Creative | Auto-generated headline from the page, your description | Model-generated headlines and descriptions, more variation |
Targeting control | Page feeds, category targets, URL rules | Fewer explicit targeting levers; more weight on exclusions |
Where you steer it | Which pages you expose | Negative keywords, brand controls, feeds and conversion signals |
Creation after Sept 2026 | Not possible — UI, Editor and API all blocked | The only option |
That fifth row is the practical heart of it. Under DSA you steered by choosing which pages the system could see. Under AI Max you steer mostly by telling it where not to go. Control moves from inclusion to exclusion, and exclusion is a maintenance activity rather than a setup activity.
What Control Do You Actually Lose?
The reflexive reaction to any automation upgrade is that you lose everything. That is not accurate, and treating it that way leads to the wrong preparation.
Control | Status | What to do about it |
|---|---|---|
Page-level targeting rules | Reduced | Move the intent into exclusions and feed hygiene instead |
Precise match-type steering | Reduced | Expect broader matching; budget for more query review |
Headline wording | Reduced | Fix it at the source — page titles and on-page copy now drive creative |
Negative keywords | Retained, and more important | This becomes your primary steering wheel. Treat it as ongoing work |
Brand exclusions | Retained | Set deliberately, or brand traffic cannibalises the campaign's numbers |
Conversion signals | Retained, and more important | The model optimises toward whatever you tell it success is. Audit that first |
Budget and bid strategy | Retained | Unchanged levers, but they act on a wider query set now |
Key Takeaway: You are not losing control so much as being handed a different set of controls. The levers that survive — negatives, brand exclusions, conversion signals and feed quality — matter more after migration than they did before, because they are now the only things standing between a much wider net and your budget.
The Negative Keyword Problem Gets Bigger
This is the single most predictable failure mode of the migration, and it is worth its own section because it is where most of the wasted spend will land.
A wider net catches more of everything, including more irrelevant queries. Under DSA, your page content was a natural boundary — the system could only match things your site plausibly talked about. AI Max reasons about intent, which means it will match queries your pages never mention, some of which will be excellent and some of which will be waste.
The negative keyword list you built for DSA was sized for a narrower problem. It will be inadequate almost immediately, and the failure is quiet: spend drifts to junk queries, blended performance sags a few percent, and nothing in the interface flags it as a distinct event. You find it weeks later while looking for something else.
The mitigation is straightforward but relentless. Review search terms far more often than you used to, especially in the first six weeks. Look for query themes rather than individual terms — you are hunting categories of irrelevance, not one-off oddities. And front-load a defensive list before migration rather than reacting after.
This is the part that suits automation well, because it is high-volume, repetitive pattern recognition on data you already have. An AI with live access to your search term report can surface the themes in one pass rather than you scrolling a 4,000-row export. If you have Google Ads connected to Claude or ChatGPT, "group last week's search terms into themes and flag the ones with spend and no conversions" is a weekly five-minute job instead of an afternoon.
Honest limitation: Aggressive negative keyword lists cause their own damage. Every negative you add is a query you will never serve on again, and broad negatives applied hastily during a migration panic can strangle a campaign's ability to find the good queries AI Max is genuinely better at surfacing. Add phrase-level negatives for clear themes; be slow with broad ones. If performance drops after a big negative push, that is a candidate cause worth checking before you blame the migration.
What Is the Migration Checklist?
Five steps, in this order: Audit → Harvest → Fence → Baseline → Migrate. The first four all happen before you flip anything.
Audit your conversion signals first. AI Max optimises harder toward whatever you have defined as a conversion. If a low-value micro-conversion is still counted as primary, a more aggressive model will chase it more effectively than DSA ever did. Fix this before anything else — it is the highest-leverage thing on the list and the one people skip.
Harvest your winners. Export the search terms your DSA campaigns actually converted on over the last 90 days. Those queries represent proven intent. Some deserve promotion to explicit keywords in a standard campaign so you keep direct control of your best traffic rather than handing all of it to automation.
Fence the account. Build the defensive negative list before migration, not after. Start from the junk themes already visible in your DSA search terms, then add the obvious category exclusions your pages happen to have shielded you from: jobs and careers, free and DIY variants, competitor names you do not want to bid on, and anything geographically irrelevant. Set brand exclusions deliberately.
Baseline everything. Record CPA, ROAS, conversion rate, impression share and query volume for the 30 days before migration, per campaign. Without a baseline you will be arguing about whether performance changed rather than by how much, and post-migration you will have no way to separate a real regression from normal variance.
Migrate deliberately, one campaign at a time. If you have several DSA campaigns, do not let them all convert on the same day. Stagger them so you can attribute any change to a specific migration rather than to everything at once.
Two things to watch in the first month: query volume, which will rise and should be expected to, and the share of spend going to queries with no conversion history. The second is your early warning that the net is catching the wrong things.
What This Means If You Run AI on Your Account
There is a wrinkle worth flagging for anyone using an AI assistant against Google Ads. The official Google Ads MCP server is read-only — it exposes three tools and cannot modify campaigns, bids or negatives. So it can help you find the junk queries but cannot act on them.
That is fine if you are happy applying negatives by hand. It is a real constraint if you were hoping to automate the loop, and it is worth understanding before you plan a workflow around it — because the capability differs sharply from one platform's official server to the next.
Read this next → Why Not Just Use the Platform's Own MCP Server?
Amazon and Meta's official servers let an AI change campaigns. Google's and Microsoft's do not. Here is exactly what each one will let a model touch, and the four gaps none of them close.
What Should You Take from This?
Dynamic Search Ads creation is blocked from September 2026 across the UI, Editor and API — API-driven campaign builds are the most exposed and should be checked now.
AI Max casts a substantially wider net, which shifts your control from inclusion to exclusion. Negatives become the primary steering wheel rather than a hygiene task.
Audit conversion signals before migrating. A more capable model optimising toward a badly defined goal is worse than a less capable one, not better.
Baseline 30 days of performance per campaign and stagger the migrations, or you will not be able to tell what changed.
Watching a Migration Across Several Accounts?
Connect Google Ads to Claude, ChatGPT or Brandlio's agent and ask for search term themes, baseline comparisons and anomaly flags across every account in one prompt — instead of one export at a time. See pricing or connect your first account.
Frequently Asked Questions
When exactly is Dynamic Search Ads being deprecated?
The upgrade into AI Max for Search begins in September 2026, at which point new Dynamic Search Ads campaigns can no longer be created through the Google Ads interface, Google Ads Editor or the API. AI Max for Search exited beta on 15 April 2026 with a migration path reaching all advertisers by September. Verify current dates in Google Ads Help, as timelines have shifted before.
Will my existing DSA campaigns stop running?
They are being upgraded rather than switched off. Existing Dynamic Search Ads campaigns are converted into AI Max for Search rather than paused, so traffic should continue. What you lose is the ability to create new DSA campaigns and some of the targeting controls DSA offered.
Can I opt out of the AI Max migration?
No. This is a platform-level deprecation with a migration path applied to all advertisers, not an optional beta. The realistic choice is whether you migrate deliberately with preparation, or have the deadline migrate you without it.
What happens to my API scripts that create DSA campaigns?
They will start failing once creation is blocked, because the block applies to the API as well as the interface and Editor. If you build campaigns programmatically this is the most urgent item on your list — audit any script or tool that creates DSA campaigns and plan its replacement before September.
Why do I need more negative keywords with AI Max?
Because AI Max casts a wider net than DSA by design. Under DSA your page content acted as a natural boundary on which queries could match. AI Max reasons about intent and will match queries your pages never mention, so exclusions replace page targeting as your main way of steering the campaign.
Should I move my best DSA search terms to exact match keywords?
Often yes. Search terms that converted consistently in DSA represent proven intent, and promoting them into a standard campaign as explicit keywords keeps direct control of your highest-value traffic rather than handing all of it to automation. Harvest them from the last 90 days before you migrate.
How long should I wait before judging AI Max performance?
Give it a full learning period and at least 30 days of stable data, and compare against a baseline you recorded before migration. Judging in the first fortnight tends to measure the learning phase rather than the steady state. Staggering campaign migrations also helps, since it lets you attribute changes to a specific campaign instead of everything at once.
Can an AI assistant apply negative keywords to Google Ads for me?
Not through Google's own MCP server, which is read-only and exposes three tools for reporting only. It can identify wasteful queries but cannot act on them. Applying negatives automatically requires a write-capable third-party server, and capability varies considerably between providers — ask for the explicit list of write tools rather than relying on marketing language.
Technical marketing specialist pushing boundaries in Google Ads, automation, and AI-driven growth systems. Paragliding and adventure enthusiast.
WordPress
