# DSAs Are Sunsetting: Your AI Max Migration Plan

DSA campaigns auto-upgrade to AI Max on Google's schedule whether you plan or not. Migrating deliberately keeps your URL controls and data intact.

Blog · Published 2026-08-10, updated 2026-10-26.

> Dynamic Search Ads are being replaced by AI Max on a published schedule: voluntary migration tools shipped in April 2026, DSA creation ends in January 2027, and remaining campaigns auto-upgrade from February 2027 — a deadline Google already moved once, from September 2026. Migrating deliberately inside the voluntary window preserves your URL rules, negatives and learning on your terms; doing nothing means the same conversion happens in February's bulk queue, configured by defaults.

Dynamic Search Ads are ending, and unusually for a Google deprecation, the
ending has a published schedule and a pushed-back deadline — which means
you currently hold the thing deprecations rarely offer: time to migrate on
your own terms. Use it. A DSA campaign that reaches February 2027
untouched gets converted anyway, in bulk, configured by defaults instead
of by you.

## The official timeline, as it stands

| Date | What happens |
| --- | --- |
| Apr 2026 | AI Max reaches GA; voluntary DSA upgrade tools begin rolling out |
| Jun 15, 2026 | DSA creation restored for the transition window |
| Sep 2026 | Auto-created assets and campaign-level broad match fold into AI Max |
| Jan 2027 | Ability to create DSAs removed |
| Feb 2027 | Remaining active DSA campaigns begin auto-upgrading |

Context worth keeping: the auto-upgrade was originally scheduled for
September 2026 and moved to February 2027 after advertiser pushback —
Google explicitly sparing accounts a forced migration during Q4. Read
that two ways at once: the dates can shift again, and the direction never
does. A delay is breathing room, not a reprieve, and treating it as a
cancellation is how accounts end up in the bulk queue. Plan against the
schedule; verify it quarterly. The broader 2026
picture sits in [what actually changed this year](https://adcopilot.cloud/blog/google-ads-in-2026-what-changed).

## What DSAs did that AI Max does differently

DSAs were a bargain with a specific shape: point Google at your site or
page feed, and it matched queries to landing pages and wrote headlines,
inside a campaign type built for exactly that. [AI Max](https://adcopilot.cloud/glossary/ai-max-for-search)
absorbs the idea — keywordless matching driven by your landing pages —
but repackages it as a feature suite on Search campaigns, alongside
keywords rather than instead of them.

Three differences decide your migration work. **Targeting grammar:** DSA
targets — page feeds, categories, URL rules — become AI Max's URL
inclusion and exclusion controls, and the mapping is where mistakes hide.
**Creative control:** DSA's generated headlines become AI Max text
customisation, steerable with text guidelines you now need to write.
**Blast radius:** DSA was contained in its own campaigns; AI Max features
attach to Search campaigns you may also run keywords in, so a sloppy
migration can leak into structures DSAs never touched.

Reporting shifts with the mechanism. DSA reporting was organised around
its own grammar — search terms joined to the landing pages Google chose —
which made "which pages earn paid traffic" a one-report question. Under
AI Max the same information exists but arrives through the campaign's
general reporting surfaces, alongside keyword traffic. Plan for the
first month of readouts to need new prompts rather than old habits, and
keep the exported DSA baseline as the translation key between the two
worlds.

## Inventory first: find every DSA dependency

Migration effort is proportional to what your DSAs actually do, which
most accounts have not looked at in years. One prompt builds the
inventory:

> Find every DSA campaign and ad group in the account. For each: spend,
> conversions and cost per conversion over the last 12 months; the page
> feeds or dynamic ad targets in use; every URL exclusion; attached
> negative keyword lists; and the top 50 search terms with conversions.
> Flag DSA queries that overlap keywords elsewhere in the account, and
> export the whole thing as my pre-migration baseline.

The overlap flag matters most. Years of DSA harvesting usually conceal a
stable core of converting queries that deserve to become
[real keywords](https://adcopilot.cloud/blog/keyword-research-from-your-own-data) regardless of
what happens to the campaign type — promoting them first shrinks what the
migration has to get right. The inventory also sizes the project honestly:
an account where DSAs are a low-spend long-tail catcher can migrate in an
afternoon, while one where DSAs quietly became the top conversion driver —
more common than owners expect — deserves the full staged treatment and a
sceptical look at why the keyword coverage never caught up.

## The deliberate migration, in order

1. **Export the baseline.** The inventory above, saved outside the
   account. Post-migration comparisons need a fixed point.
2. **Promote the proven core.** Converting DSA queries become exact or
   phrase keywords in the right ad groups, with ads written against them.
3. **Rebuild the fences.** URL exclusions carry your intent: careers,
   support, out-of-stock and legal pages must be excluded in AI Max
   before it can route traffic. Same for negative lists — attach, do not
   assume.
4. **Write text guidelines.** DSA generated headlines with no brief from
   you; AI Max at least accepts instructions. Banned claims, required
   qualifiers, tone — write them before the system writes anything.
5. **Upgrade one campaign in the voluntary window.** Use the in-product
   upgrade path on your mid-stakes DSA campaign first — not the biggest,
   not a token — and run it against the baseline for 30 days.
6. **Then do the rest,** applying what the first told you, finishing
   before January turns the queue into February's default conversion.

## The first 30 days post-upgrade: what to watch

AI Max succeeds or embarrasses in three places, so the watchlist is
short and weekly:

- **Query drift** — expanded matching versus your DSA-era terms; the
  [weekly policing prompt](https://adcopilot.cloud/blog/ai-max-for-search-opt-in-decision) with
  drafted negatives applies unchanged.
- **Destination drift** — which URLs traffic actually lands on, checked
  against your exclusion intent; URL expansion is the feature most likely
  to surprise.
- **Economics against baseline** — cost per conversion versus the
  exported DSA numbers, judged at 30 days, not at 3.

Each check is a read; each fix — a negative, an exclusion, a guideline
tightened — is one write, which AdCopilot proposes first, and your AI client asks before it runs if it is set to ask before AdCopilot's write tools. An account that runs this loop owns
its upgrade.

## If you do nothing: what the forced upgrade changes

The do-nothing scenario is worth stating concretely, because it is the
default and defaults deserve prices. In February 2027 your remaining DSA
campaigns convert on Google's mapping, configured by the migration's
defaults rather than your decisions. Nobody exports a baseline, so the
before-and-after argument is unwinnable either way. Your URL exclusions
and negatives carry only as well as the automated mapping carries them —
probably fine, verified by no one. Text runs on default behaviour until
someone notices a headline worth objecting to. And all of it lands in the
month after Q4, when attention is scarcest and the year's comparisons are
already muddy.

None of that is catastrophe; it is compounded carelessness — the same
features you would have adopted anyway, minus the fences, the baseline
and the first month of watching. Given that the watching now costs a
weekly prompt, the do-nothing discount is no longer worth what it used
to be.

If your account still runs DSAs, the decision has been made for you; only
the manner is left, and the manner is worth real money: the deliberate
path costs a few working sessions spread over a quarter, while the
February path costs whatever the defaults get wrong in the month nobody
was watching. Start the inventory this week, migrate inside the window —
and if AI Max itself is still an open question for your non-DSA
campaigns, the [opt-in decision framework](https://adcopilot.cloud/blog/ai-max-for-search-opt-in-decision)
is the companion read.

## Questions

### When exactly do DSAs stop working?

Per Google's announced timeline: new DSA creation — restored in June 2026 for the transition — is removed in January 2027, and from February 2027 remaining active DSA campaigns are upgraded automatically. Two related transitions kept their September 2026 date: automatically created assets and the campaign-level broad match setting both fold into AI Max. These dates have shifted once already — the auto-upgrade was originally September 2026 — so verify against Google's current announcements before you schedule around them.

### Will my DSA query history carry over?

Your historical reporting survives — past DSA performance stays queryable against the old campaign structure — but continuity forward does not. Post-migration, matching runs through AI Max's mechanisms, so query-level history restarts in practice and comparisons across the boundary are apples-to-oranges. That is the strongest argument for exporting a pre-migration baseline: top search terms, landing pages, cost per conversion — the numbers you will want when judging whether the upgrade helped.

### Should I migrate to AI Max or rebuild as standard keyword campaigns?

Depends on what your DSAs were doing. If they mostly harvested long-tail queries against a large changing site, AI Max is the designed successor — keywordless matching from your landing pages continues under new controls. If years of DSA search terms show a stable core of converting queries, consider promoting that core into keyword campaigns first, then let AI Max hunt the remainder. Many accounts land on both: harvested keywords for the proven, AI Max for the frontier.

---

This is the markdown copy of https://adcopilot.cloud/blog/dsa-to-ai-max-migration. The index of the site's pages is https://adcopilot.cloud/llms.txt.
