Digital Peax/Journal/Performance
Performance

AI Max lands on September 1, and the two migration paths are not the same

·8 min read·
AI Max lands on September 1, and the two migration paths are not the same

Google will start automatically upgrading Search campaigns to AI Max between September 1 and 30. We flagged the date two weeks ago. The developers blog has since confirmed the details, and the one worth writing down is that the two migration cohorts land on different defaults. The audit you need to run is account-specific, and part of the change already happened on August 3.

The Google Ads Developers Blog confirms two cohorts are in scope this autumn: campaigns using the campaign-level broad match setting, and campaigns using standalone Automatically Created Assets. Dynamic Search Ads were originally in the same wave, but Google pushed them to February 2027, with in-account notices during September and reminders from January 15. If you ran the audit we described when the September 1 date was announced, you already have the two lists. What follows is what happens to each of them.

The part that already happened

Since August 3 you cannot create new campaign-level broad match configurations or new legacy ACA setups. Not in the UI, not in Google Ads Editor, not in any version of the API. A lot of accounts have not noticed, because nothing existing broke. But if you have a campaign build template, a naming-convention script, or a client onboarding checklist that assumes those toggles exist, it is already broken.

The two cohorts get different defaults

This is the detail worth writing down, because the risk profile is not equal.

Campaigns migrating from campaign-level broad match land with search term matching on, text customization off, and final URL expansion off. Existing brand inclusion and exclusion lists carry over. Functionally this is close to what those campaigns already did. One practitioner quoted by PPC Land called it "basically just a rebrand and a shove into a new bucket," which is a fair read for accounts that were already running broad match with a solid negative list.

Campaigns migrating from standalone Automatically Created Assets land with search term matching on and text customization on. That is the one to look at. Per PPC Land's description of the feature, text customization combines extractive techniques, pulling snippets from landing page titles and meta tags, with generative models grounded in page content. Assets refresh roughly every 48 hours, and machine-written headlines serve when Google predicts they will outperform yours. Pinned assets in responsive search ads are still respected, which is the main lever you have if specific copy is non-negotiable.

How AI Max text customization works: it pulls snippets from landing page titles and meta tags, generates copy grounded in page content, refreshes assets roughly every 48 hours, and serves machine-written headlines when they are predicted to outperform yours, while pinned responsive search ad assets are still respected
What the ACA cohort inherits: a system rewriting ad copy on a two-day cycle, with pinned assets as the one lever that always holds.

So the practical difference: one cohort gets broader query eligibility it mostly already had. The other gets broader query eligibility plus a system rewriting ad copy on a two-day cycle. If you run regulated clients, trademark-sensitive copy, or anything where a claim in a headline creates a compliance problem, the ACA cohort is where you spend your audit time.

What Google is claiming, and how much weight to put on it

Google's blog post says AI Max for Search campaigns see an average of 7% more conversions or conversion value at a similar CPA or ROAS when using the full feature suite compared with limited feature usage. Read the qualifiers carefully. This is internal 2026 data, it excludes retail advertisers, and the comparison is full feature adoption against partial adoption inside AI Max, not AI Max against a well-built manual setup. It is a reason to test, not a reason to skip the audit.

The controls you still have, and what they are not

Search term matching can be disabled at ad group level. Text guidelines let you exclude terms, capped at 25 exclusions. AI Brief, added on April 30, takes natural-language instructions for steering the system. Brand inclusions and exclusions carry through the migration. Campaigns using final URL expansion carry a mandatory text disclaimer requirement.

Twenty-five excluded terms is not much for a multi-brand account. Budget for that constraint now rather than discovering it in October.

Worth being clear on what these controls are not. None of them give you the pre-migration setup back. There is no opt-out documented in either Google's developer post or its blog announcement, and the DSA path only bought a delay, not an exemption. The realistic goal for September is not avoiding the change, it is arriving at it with a baseline, a pinned-copy audit, and an exclusion list already in place.

If you manage accounts through the API

The migration writes timestamps you can query. The developers blog gives the fields directly: campaign.aca_migration_date_time and campaign.broad_match_migration_date_time. Query those during September and you get a clean list of what actually moved and when, rather than trusting an in-account notification you may not see. New API versions released after September 1 remove the legacy entities entirely, and older versions run to their normal sunset, typically September 2027.

Two automation changes in the same month

One compounding factor for anyone running multi-language accounts: Google is also removing campaign-level language targeting from Search campaigns in late September, replaced by automatic matching on ad language and user signals. Performance Max keeps language settings for YouTube, Display, Discover and Gmail, losing them only for Search. Two automation changes to Search targeting inside the same month makes attributing a performance shift to either one considerably harder. If you can, note the exact dates each lands per account so you have something to segment against later.

Also worth noting this week

Where the risk actually sits: the broad match cohort mostly gets a rename. The ACA cohort gets its ad copy rewritten on a 48-hour cycle, so the pinned-asset audit and the 25-term exclusion list are September's real work.

What to do this week

  • Segment your accounts into the two cohorts. Filter Search campaigns by campaign-level broad match and by standalone automatically created assets. Treat them as separate lists, because they get different defaults and carry different risk.
  • Pull a 90-day performance baseline for both lists before September 1. Impressions, CTR, CPA or ROAS, and search term composition. After migration you will need a pre-change reference that is not contaminated by the language targeting change landing later in the month.
  • Audit ad copy in the ACA cohort for anything that cannot be rewritten. Prices, guarantees, regulatory language, trademarked phrasing. Pin those assets now. Do not assume you will get a warning before text customization starts serving.
  • Build your text guideline exclusion list against the 25-term cap. Competitor names, discontinued products, and claims you are not allowed to make. Prioritize, because you will run out of slots.
  • If you use the API, schedule a September query on the two migration timestamp fields so you have a dated record of exactly which campaigns moved.

September will sort accounts into the ones that chose their AI Max configuration and the ones that inherited it. Getting client accounts into the first group, with the baseline to prove what changed, is the job of our performance marketing services.


Sources

The Peax Brief

One sharp idea on data-driven growth, every other week. No spam, unsubscribe anytime.

$
Book a strategy call

Ready to grow on purpose?

Tell us where growth is leaking and we’ll map exactly what we’d do about it. A free, no-obligation strategy session with the people who’d run your account.

Languages
English · Turkish
peax://new-enquiry