Between September 1 and 30, 2026, Google automatically upgrades two kinds of Search campaigns to AI Max: those using the campaign-level broad match setting, and those using Automatically Created Assets. AI Max is not a new campaign type, so your campaign IDs survive. What changes is the reporting layer underneath them.

Most coverage of this migration is written for the person clicking through the Google Ads UI. This one is written for whoever owns the pipeline that carries Google Ads data into a warehouse or a BI tool, because that is where the upgrade actually lands: new segments, a new report view, and one silent change to landing-page URLs that breaks joins if nobody is watching. Every claim below was checked against Google's own documentation on September 2, 2026.

Key Takeaways

  • The September 1-30 window applies to campaign-level broad match and Automatically Created Assets campaigns only. Dynamic Search Ads were moved out to February 2027.
  • AI Max is a setting on an existing Search campaign, not a new campaign type. There is no campaign-type migration to reconcile in your warehouse.
  • Three reporting surfaces are new in the API: the segments.search_term_match_source segment, the ai_max_search_term_ad_combination_view view, and expanded_landing_page_view.
  • Final URL expansion means the URL a person landed on is not always the final URL stored on the ad. Any join keyed on that URL degrades quietly.
  • Since August 3, 2026 you cannot create new campaign-level broad match or legacy ACA campaigns, so the affected set is fixed and countable today.
  • Opting out is per feature and takes effect in campaign settings, but the practical deadline is the start of the migration window.

Is AI Max a new campaign type?

No. Google's documentation is explicit: "AI Max is not a new campaign type. It is an optimization layer that you activate within your existing Search campaigns." The upgrade flips settings on campaigns you already have, which means campaign IDs, campaign names and the rest of your campaign structure carry through the migration rather than being replaced.

That is the single most useful fact for anyone maintaining a reporting pipeline, and it is the one nobody writing UI walkthroughs bothers to state. There is no new campaign type or subtype to add to a data model, no second campaign object to union with the first, and no obvious break in a campaign's time series on September 1. Google describes the design goal as letting advertisers "maintain familiar Search campaign structures while upgrading specific settings like brand controls and broad match into a single, AI-driven framework."

One honest caveat: Google's help documentation states what happens to campaign structure, but it does not publish an explicit statement about how historical rows are handled in reporting. Structure continuity is documented; a guarantee about historical data is not. Treat the first as sourced and the second as something to confirm against your own exports in the first week of October.

What is actually being upgraded in September, and what is not

The migration has been announced in stages, and the stages have moved, so the dates are worth stating precisely.

  • August 3, 2026. Google stopped allowing the creation of new campaign-level broad match and legacy Automatically Created Assets campaigns across the UI, Google Ads Editor and the API, per Search Engine Land's timeline. The practical effect for reporting is that the population of affected campaigns is frozen and can be enumerated today.
  • September 1-30, 2026. Automatic migration of those campaigns to AI Max. Google's own post states that "campaigns using Automatically Created Assets (ACA) and campaign-level broad match setting will continue to be auto-upgraded starting in September 2026."
  • February 2027. Dynamic Search Ads. Google is "extending the timeline for Dynamic Search Ads sunset and auto-upgrade, which will begin in February 2027".

The DSA date is worth a second look, because it is a revision rather than an original plan. Google announced the DSA delay to February 2027 on its Ads Developer Blog in June 2026, having previously placed DSA on a nearer track. If you built a Q4 plan on the earlier date, the plan is wrong in your favour, and DSA campaigns are not in scope for anything that happens this month.

Between those dates, Search Engine Land's timeline notes in-account notices encouraging voluntary DSA upgrades starting September 2026, reminder notifications on January 15, 2027, and automatic DSA migration running February 1-28, 2027.

Which reporting fields change

This is the part that has no equivalent in the UI-focused coverage. Google's AI Max reporting documentation for the Google Ads API names the new objects directly.

A new segment that tells you where a search term came from. segments.search_term_match_source attaches to the search term view and takes three values:

  • ADVERTISER_PROVIDED_KEYWORD, the traffic your keywords earned.
  • AI_MAX_BROAD_MATCH, traffic AI Max expanded out of an existing keyword.
  • AI_MAX_KEYWORDLESS, traffic matched from page and asset content with no keyword involved at all.

That third value is the interesting one. It is the first time a mainstream Search reporting surface labels traffic that originated from no keyword, and if your models bucket spend by keyword, it is the bucket that will start filling with nulls.

A new view for the most granular cut. ai_max_search_term_ad_combination_view joins the search term, the landing page and the headline for each combination, which is the level at which AI Max actually makes decisions.

Two warnings attached to that view matter more than the view itself, because both produce numbers that look valid and are not. It overlaps the classic search term report: Google states the two "are not mutually exclusive", that "many search terms generated by AI targeting are reported in both views", and the instruction is blunt, "never sum metrics from these two views". Union them in a model and you overstate clicks, with the combined total exceeding campaign clicks as the tell. The opposite error exists too: "some search terms that do not have enough query activity are omitted from the search terms report to maintain our standards on data privacy", so term-level rows legitimately fall short of campaign totals. Reconcile to the campaign, never to the sum of the two term-level views.

A new landing-page view. expanded_landing_page_view, carrying expanded_final_url, exposes the URL that was actually served rather than the one configured on the ad.

A campaign-level flag. campaign.ai_max_setting.enable_ai_max tells you, per campaign, whether AI Max is on. This is the field to pull first, because it turns "which of our campaigns got upgraded" into a query instead of an audit.

Google's documentation gives a working query shape for the search-term cut:

SELECT
  campaign.id,
  campaign.name,
  search_term_view.search_term,
  segments.search_term_match_source,
  metrics.impressions,
  metrics.clicks,
  metrics.conversions
FROM search_term_view
WHERE
  segments.search_term_match_source IN ('AI_MAX_KEYWORDLESS', 'AI_MAX_BROAD_MATCH')
  AND segments.date DURING LAST_30_DAYS

One deliberate omission: the documentation page does not state which Google Ads API version first carries these fields. Rather than guess at a version number, check the release notes for the version your client library is pinned to before you promise a date to anyone.

And in the interface

Google's reporting help page for AI Max documents the matching set of UI changes, which matter if anyone on the team reconciles exports against screenshots:

  • The search terms report gains a "Search terms and landing pages from AI Max" view, an "AI Max" value for the match type filter, and a "Source" column you have to add to the default view yourself.
  • The keywords report gains two aggregate rows at the bottom: "Total: AI Max expanded matches" and "Total: AI Max landing page matches".
  • The landing pages report gains a "Selected by" column, which per Google "will indicate if the URL was selected by AI Max".
  • The asset report gains an "Expanded final URL assets" tab, and text assets Google generated are marked "Google AI" in the "Added by" column.

There is one reconciliation trap documented on that same page. Google warns that "when filtering for match type = 'AI Max' the search terms report may show lower numbers for AI Max. This filter doesn't consider 'Other search terms'." If someone compares that filtered UI number against an API pull segmented by match source, the two will disagree, and the UI is the one leaving rows out.

The join that quietly breaks

Final URL expansion is a campaign-level control that ships as part of AI Max. Google describes it as sending "traffic to the most relevant URLs on your website when it's likely to result in better performance". Read that as a reporting statement and the consequence is immediate: the URL stored on the ad is no longer reliably the URL the visitor reached.

Most marketing data models lean on that URL more heavily than anyone remembers. Landing-page URLs get parsed for UTM parameters, grouped by path to roll pages into content categories, joined against a CMS export to attach page metadata, and matched against on-site analytics to stitch a session to a click. Every one of those joins assumes the ad's final URL is the page that was served.

When that assumption breaks, it does not throw an error. Rows do not disappear. They land in an "other" bucket, or match the wrong page group, or silently fail to join and quietly reduce a category's measured performance. This is the failure mode most worth pre-empting, precisely because nothing alerts on it.

The fix is not clever, it is just early: read expanded_final_url from expanded_landing_page_view, and treat the ad's configured final URL as intent rather than fact. If you maintain a page-grouping layer, key it on the served URL and keep the configured URL as a separate column so you can measure how often the two diverge. That divergence rate is itself a useful number to have by October.

How to opt out, and by when

Opting out is per feature, not all or nothing. Per Google's setup documentation:

  1. Open Campaigns and go to the Settings tab.
  2. Check the box next to the campaigns you want to change.
  3. Open the AI Max section of campaign settings.
  4. Toggle AI Max off entirely, or turn off individual features. Note the two-level control on search term matching: it "must be turned on at the campaign level in order to be used at the ad group level", and at ad group level "it can be turned on or off". The campaign switch is the master and the ad group switch only narrows it, so an opt-out done purely in ad groups leaves the campaign setting on. Text customization and final URL expansion are campaign-level controls.
  5. Re-enabling later restores the feature-level settings you had before, so a temporary opt-out is not destructive.

On the deadline, it is worth being precise about what is documented and what is not. Google's help pages do not themselves publish a hard opt-out cutoff date. What is documented is the August 3 creation freeze and the September 1-30 migration window, so the effective deadline is the start of that window rather than a separately announced cutoff. If you see a specific date quoted as an official Google deadline, check whether it traces to a Google page or to a summary of one.

Finding every affected campaign before the window closes

The advice written for single-account operators is to look through campaign settings. That does not survive contact with fifty accounts under an MCC, and it is the gap in every other article on this migration.

Because the affected population was frozen on August 3, it can be enumerated rather than hunted. A workable sequence:

  1. Pull campaign.ai_max_setting.enable_ai_max for every campaign across every account, and snapshot it now, before the window ends. Without a pre-migration snapshot you lose the ability to say which campaigns were changed by Google rather than by a person.
  2. Re-pull the same field weekly through September. The diff is your migration log, and it is the only version of that log you will have.
  3. Segment the search terms report by segments.search_term_match_source from day one, so that the moment keywordless traffic appears you can size it rather than discover it in a quarterly review.
  4. Start recording expanded_final_url alongside your existing landing-page field, and monitor how often the two differ per campaign.
  5. Decide the opt-out question per campaign before the end of September, and record the decision somewhere durable. In three months, "did we choose this or did it happen to us" is a question someone will ask.

How Improvado fits

Improvado's role here is the boring, useful one: it pulls Google Ads reporting into a warehouse alongside your other paid channels, so a platform-side change like this one shows up in a single pipeline rather than in every spreadsheet that touches paid search. That matters most in exactly the situation this migration creates, where the question is not "what happened in this campaign" but "which of our two hundred campaigns changed shape last week, and what did that do to the numbers we report upward".

If you already centralize Google Ads data, the work this month is to confirm that the new fields are in your extract before the window closes, and that your landing-page logic reads the served URL. If you do not centralize it yet, this migration is a reasonable forcing function, because the alternative is reconciling the change by hand in each account.

Frequently Asked Questions

Is AI Max a new campaign type in Google Ads?

No. Google states that AI Max "is not a new campaign type" but "an optimization layer that you activate within your existing Search campaigns". Campaign structure carries through the upgrade, so there is no new campaign type to add to a reporting data model.

Which campaigns are auto-upgraded in September 2026?

Campaigns using the campaign-level broad match setting and campaigns using Automatically Created Assets. The migration runs from September 1 to 30, 2026. Dynamic Search Ads are not included in this window.

When are Dynamic Search Ads upgraded to AI Max?

February 2027. Google extended the Dynamic Search Ads sunset and auto-upgrade timeline, with in-account notices from September 2026, reminders on January 15, 2027, and automatic migration during February 1-28, 2027.

How does AI Max change the Google Ads API reporting schema?

It adds the segments.search_term_match_source segment with three values, the ai_max_search_term_ad_combination_view report view, the expanded_landing_page_view view carrying expanded_final_url, and the campaign.ai_max_setting.enable_ai_max campaign field.

Will my search terms report still show the same data?

Not exactly. Search terms now carry a source value distinguishing keyword-driven traffic from AI Max broad match and keywordless matching. Google also warns that filtering the report by match type "AI Max" can show lower numbers, because that filter excludes "Other search terms".

Can I turn off final URL expansion without turning off AI Max?

Yes. Text customization and final URL expansion are both campaign-level controls and each can be switched off on its own. Search term matching is the one to watch, because it works across two levels: Google requires it to be on at campaign level before it can be used at ad group level, where it can then be turned on or off per ad group.

What breaks in a marketing data pipeline when AI Max turns on?

The most common breakage is any join keyed on the ad's final URL. Final URL expansion can serve a different page than the one configured, so UTM parsing, path-based page grouping and CMS joins can silently mismatch unless the pipeline reads the expanded final URL instead.