Google's DSA-to-AI Max migration just got a reprieve: on June 11, 2026, Google Ads Liaison Ginny Marvin announced the automatic migration of Dynamic Search Ads to AI Max is postponed from September 2026 to February 2027, citing advertiser feedback and the need to avoid disrupting Q4 planning. For most PPC teams that read like a problem deferred. It is closer to a problem clarified.
The headline is real, but the nuance is where the money is. The delay applies only to DSA automigration. Campaigns running Automatically Created Assets (ACA) and campaign-level broad match settings still migrate to AI Max on the original September 2026 schedule — that timeline was not changed. So the reprieve is narrower than the coverage suggests, and it lands unevenly across your account.
This playbook lays out what actually changed, the two separate migration clocks now running in parallel, what AI Max really is under the hood, how Google's own performance claims line up against independent data, and the single highest-leverage thing to do with the extra runway: run a controlled Campaign Experiment as soon as DSA creation reopens around June 15, so February 2027 doesn't catch you migrating blind.
- 01The DSA delay is real but narrow.Automatic DSA-to-AI Max migration moves from September 2026 to February 2027, per Google's June 11 announcement. Ginny Marvin attributed it to advertiser feedback and avoiding Q4 disruption.
- 02ACA and broad match still migrate in September 2026.The September timeline for Automatically Created Assets and campaign-level broad match settings was not changed. Most Search accounts are affected by this in roughly three months — regardless of the DSA delay.
- 03AI Max is three features, not a campaign type.Smart Search Term Matching, Text Customization, and Final URL Expansion get layered onto standard Search campaigns. AI Max is now the default for newly created Search campaigns as of the April 2026 GA.
- 04Google's headline number is not a DSA comparison.The often-quoted +7% is full-suite AI Max versus matching-only AI Max, on non-retail accounts — not AI Max versus DSA. Independent retail data shows +13% revenue but +16% higher CPA.
- 05Use the window to baseline, not to wait.New DSA creation is set to reopen around June 15, 2026, so you can run a clean Campaign Experiment as soon as it returns and read real output before February 2027 forces the switch. Setting text guidelines is a separate, low-effort protective move.
01 — What ChangedA five-month reprieve, announced June 11.
On June 11, 2026, Google posted the change to its Ads Developer Blog and Ginny Marvin, the Google Ads Liaison, confirmed it publicly: the DSA automigration deadline shifts from September 2026 to February 2027. The stated reason was advertiser feedback and the goal of not interfering with the busy Q4 period. Alongside the delay, Google said new DSA campaign creation — which the original sunset plan had been set to block — would be restored as of June 15, 2026.
For context, the original DSA retirement was announced on April 15, 2026 with September 2026 as the hard deadline, the same week Google released voluntary upgrade tools and AI Max for Search reached general availability after roughly a year in beta. The June 11 change keeps the destination identical — every DSA campaign still ends up on AI Max — but resets the clock and reopens creation in the interim.
There is a sting in the tail. Google said the ability to create new DSA campaigns will be removed again in January 2027, immediately before the February 2027 automigration begins. So the reopening creation window is finite: it opens in mid-June and closes in January. That is the runway you actually have to build and read a clean test.
"We've heard your feedback loud and clear: you need more time to transition from Dynamic Search Ads (DSA) to AI Max."— Ginny Marvin, Google Ads Liaison, June 11, 2026
Google also said it would roll out voluntary one-click upgrade tools over the following weeks — tools designed to preserve historical reporting and minimize learning-phase disruption. When DSA campaigns are automatically migrated in February 2027, Google states legacy settings will be preserved as closely as possible, existing URL controls will carry over, and all three AI Max features will be enabled by default. That last detail matters: the default-on behaviour is exactly why a pre-migration baseline is worth having.
02 — Two ClocksThe delay is DSA-specific, not a blanket pause.
The single most under-explained point in the coverage is that there are now two migration clocks running at once. The DSA clock moved to February 2027. The ACA-and-broad-match clock did not — those campaigns still migrate to AI Max in September 2026. If you read the announcement as "the AI Max migration was delayed," you will be wrong about most of your account.
The table below lays all four tracks side by side, with the window to act counted from this article's date. Read it once for the September exposure, then again for the DSA runway — they are different problems with different deadlines.
| Campaign element | Original Sept 2026 plan | Updated (June 11 announcement) | Window to act |
|---|---|---|---|
| DSA campaigns — new creation | Blocked from Sept 2026 | Reopens ~June 15, 2026 · blocked again Jan 2027 | ~7 months |
| DSA campaigns — automigration | September 2026 | February 2027 | ~8 months |
| ACA campaigns (Automatically Created Assets) | September 2026 | September 2026 — unchanged | ~3 months |
| Campaign-level broad match settings | September 2026 | September 2026 — unchanged | ~3 months |
03 — The ProductAI Max is three features, not a new campaign type.
A common misconception is that AI Max is its own campaign type you opt into. It is not. Per Google's documentation, AI Max for Search is a suite of three features layered onto standard Search campaigns. Understanding the three is what lets you reason about where the risk lives when DSA campaigns inherit them by default.
Smart Search Term Matching
Extends reach using broad match plus keywordless technology — conceptually similar to how DSA matched queries from site content, but driven by real-time intent signals rather than static page crawls.
Text Customization
Generates headlines and descriptions in real time, drawing on your existing ads, landing-page content, and Gemini. This is the layer your text guidelines constrain — set them and the AI writes inside guardrails.
Final URL Expansion
Lets AI Max send users to any relevant page on the site rather than the specified landing page. It requires Text Customization to be active and disables pinned assets when a different URL is selected.
Two operational notes carry weight. First, AI Max is now the default setting for newly created Search campaigns as of the April 2026 GA — so even net-new work is AI Max unless you actively dial features back. Second, Final URL Expansion disabling pinned assets is a quiet but real loss of control: teams that rely on pinned headlines for compliance or brand-mandated phrasing need to know that a URL switch can override them. Neither is disqualifying. Both are reasons to test on your own traffic rather than trust the category-level numbers.
04 — The GapWhat Google claims versus what the independent data shows.
The most-quoted AI Max figure is "+7% more conversions or conversion value." It is worth reading carefully, because the baseline is not what most people assume. Per Google's own blog, that +7% is for advertisers using the full AI Max feature suite compared with using search term matching alone — in other words, AI Max-with-everything versus AI Max-with-less. It is not a comparison against DSA. Retail advertisers were explicitly excluded from that benchmark, and Google's broader non-retail figures ran higher (+14%, and +27% for accounts previously dependent on exact and phrase match).
Independent measurement tells a more textured story. The table below puts the vendor claims and the third-party data in the same frame — something the standard coverage rarely does. The unlock is the caveat column: once you see that Google's baseline is AI Max itself, the +7% stops reading as a migration promise.
| Study | Sample | Reported lift | Efficiency | Caveat |
|---|---|---|---|---|
| Google — vendor self-reports | ||||
| AI Max full suite vs matching-only | Vendor · non-retail · undisclosed | +7% conversions / value | Similar CPA / ROAS | Baseline is AI Max-only, not a DSA head-to-head |
| Non-retail broad benchmark | Vendor · undisclosed | +14% | Not stated | Retail advertisers excluded |
| Exact / phrase-heavy accounts | Vendor · undisclosed | +27% | Not stated | Selection skewed to high keyword density |
| Independent — third-party data | ||||
| Smarter Ecommerce (Mike Ryan) | Independent · retail · 250+ campaigns | +13% median revenue | +16% median CPA | ROAS spanned +42% to −35% |
| Xavier Mantica 4-month test | Independent · 1 account | Worst of all match types | $100.37 CPA vs $43.97 phrase | Single account — not generalizable |
| PPC practitioner poll (Adriaan Dekker) | Self-selected LinkedIn poll | 16% reported good performance | — | Directional signal only |
The Smarter Ecommerce analysis is the most useful single data point because it is independent, retail-specific, and large enough to matter: across 250-plus campaigns, the median result was +13% revenue paired with +16% higher CPA, and a ROAS range that swung from +42% all the way to −35%. That is not a clean win and not a clean loss — it is a real efficiency trade with wide variance, which is precisely the profile that demands testing rather than blanket adoption.
05 — The PlaySpend the runway running a Campaign Experiment.
Here is the differentiator most coverage misses. With new DSA creation set to reopen from mid-June and automigration not arriving until February 2027, you now have an unusually clean set of conditions for a controlled test — a defined window with both arms of the comparison available at the same time. The wrong response to a reprieve is to do nothing; the right one is to convert the time into baseline data you will never be able to gather once DSA is gone.
The method is three steps, and it maps directly onto the standard paid-search experiment discipline.
Restore the DSA baseline
Use the reopening creation window to keep a representative DSA campaign live and instrumented. This is your control arm and your historical reference — capture CPA, conversion volume, and ROAS before anything migrates.
Run the AI Max experiment
Stand up an AI Max test arm with the full feature suite on comparable budget and audience. Treat it like any asset experiment: equal spend, clear primary metric, enough time to clear the learning phase before you read it.
Read the output, then decide
Compare CPA and ROAS on your own traffic, not the vendor average. If AI Max holds efficiency, lean in early; if it does not, you have evidence to set guardrails before February 2027 forces the switch anyway.
If you already run structured tests inside Performance Max, this is familiar ground — the same controls apply. Our walkthrough of how to run asset experiments in Performance Max transfers almost directly to AI Max, and the broader sunset context sits in our DSA sunset agency playbook. If you are standing AI Max up for the first time, start from how to set up AI Max for Search. The point of all three is the same: get the test instrumented before the deadline removes your control.
A reprieve you spend waiting is a reprieve you wasted; a reprieve you spend measuring is the cheapest insurance you will buy this year.— Digital Applied, on the February 2027 window
06 — Protective SetupSet text guidelines now — a 20-minute hedge.
Even if you do nothing else before February 2027, there is one low-effort move that pays off the moment migration happens. Text guidelines — the brand-control layer that governs AI-generated copy — became available globally to all advertisers on February 26, 2026, covering both AI Max for Search and Performance Max. They let you set up to 25 term exclusions and up to 40 natural-language messaging restrictions per campaign, and they apply only to AI-generated assets, not to your manually written ads.
The logic is simple. When DSA campaigns automigrate, all three AI Max features turn on by default, including the Gemini-driven Text Customization. If your guidelines are already in place, the AI writes constrained copy from day one of automigration instead of free-form output you have to clean up afterward. It is a protective setup most advertisers have not bothered with — which is exactly why doing it now is cheap leverage.
07 — Your MoveWhat to do, by account type.
The right action depends on what your account actually leans on. The two clocks mean the "do nothing until 2027" option is only safe for a narrow slice of advertisers. Read down the column that matches your account and act on the nearest deadline first.
DSA is a meaningful share of Search
You got the real reprieve — eight months. Use the reopening creation window to run the DSA-vs-AI Max experiment now, and capture a baseline before January 2027 closes creation again. Don't coast.
Account relies on ACA or broad match
Your clock did not move. These campaigns migrate in September 2026, roughly three months out. Audit which campaigns are affected, set text guidelines, and decide whether to pre-empt the switch on your terms.
Keyword-driven, low DSA exposure
Lower urgency, but not zero — AI Max is already the default for new Search campaigns. Set text guidelines as a hedge and test AI Max on one campaign so the eventual shift is a decision, not a surprise.
Managing many accounts
Triage by exposure: flag every account's ACA and broad-match footprint for the September deadline first, then sequence DSA experiments across the portfolio through the reopening creation window. Standardize the test design once.
Looking past the immediate deadlines, the strategic read is that Google is steadily collapsing manual control surfaces into AI-managed ones — DSA into AI Max, manual assets into Automatically Created Assets, keyword targeting into keywordless matching. The advertisers who keep their footing through that shift will be the ones who treat each migration as a measurable hypothesis, not a mandate to accept on faith. The February 2027 date will arrive; the only question is whether you reach it holding your own data or Google's. If you want a partner to design and run that test properly, that is the core of our paid media management work.
08 — ConclusionA delay is an opportunity, if you use it.
The destination didn't change — only your time to prepare for it.
Google's decision to push DSA automigration to February 2027 is a reprieve, not a reversal. Every Dynamic Search Ads campaign still ends up on AI Max; the change is that you now have roughly eight months and a reopening creation window to get ready on your own terms. The trap is reading the headline as "deadline cancelled" and shelving the work.
The two clocks are the part to internalize. ACA and broad-match campaigns still migrate in September 2026, so for most accounts the nearest deadline is three months out, not eight. And the performance picture is genuinely mixed — independent retail data shows AI Max can lift revenue while raising CPA, with ROAS swinging both ways. That is an argument for measurement, not for trust in a vendor average.
So the move is unglamorous and high-return: restore a DSA baseline, run a controlled AI Max experiment, set text guidelines as a 20-minute hedge, and decide on your own numbers before February 2027 makes the decision for you. Treat the reprieve as a testing mandate and it becomes the cheapest insurance you will buy all year. Waste it and you migrate blind — which is the one outcome the extra runway was meant to prevent.