Home/Insights/Google moved your Search campaign to AI Max. Which of the three settings should stay on?

Google moved your Search campaign to AI Max. Which of the three settings should stay on?

AI Max is three settings, not one. A decision table, a conversion-sized read rule and a seven-step method for campaigns Google migrated in September 2026.

BY Esh, FOUNDER, ETHOS ENGINEPUBLISHED OCTOBER 5, 20268 MIN READ
KEY TAKEAWAYS
  1. The two migrated groups got different defaults. Check which of the three settings is actually on.
  2. Read the AI Max rows after 30 times target CPA in spend. Trim at 23 or fewer conversions.
  3. The AI Max label is not proof of new conversions. Check the campaign total and the marginal cost.
  4. Google's 14% and 7% figures are internal and exclude retail. Independent data shows a flat median and a wide spread.

In September 2026 Google moved a set of Search campaigns to AI Max without asking. Most guides written since then list what to check. Few say how much data you need before deciding, or that the answer differs for each of the three settings inside AI Max. This article gives a decision table, a read rule sized by conversions and a seven-step method.

What did Google change in September?

Two legacy settings were retired. One was the campaign-level broad match setting. The other was automatically created assets, which Google had already renamed text customization. Search Engine Land reported the schedule in mid-August. New use of both was blocked from August 3, and existing campaigns were migrated in place between September 1 and 30. Brand inclusions and exclusions carried over.

Dynamic Search Ads are a separate case. Google updated its announcement on June 11 to push their automatic migration back to February 2027.

AI Max is a bundle of three features:

  • Search term matching uses broad match and keywordless matching to reach queries your keyword list does not name.
  • Text customization writes extra headlines and descriptions from your ads and landing pages.
  • Final URL expansion sends a click to a different page on your domain when Google predicts it will perform better.

The two migrated groups did not get the same defaults. Google's announcement says campaigns that used automatically created assets were upgraded with search term matching and text customization on. Campaigns that used the broad match setting got search term matching only. PPC Land and Choice Marketing's dossier, which cites Google's developer blog, both report that final URL expansion was left off for the two groups. Check your own settings instead of assuming.

Do the performance numbers settle it?

They do not, and the sources disagree.

Google's launch post from May 2025 said advertisers typically see 14% more conversions or conversion value at a similar CPA or ROAS, and 27% for campaigns still mostly on exact and phrase keywords. The footnote limits this to Google internal data for non-retail advertisers. The figure quoted during the migration is smaller and narrower: 7% more for the full suite compared with search term matching alone, again internal and non-retail. Google has published no sample or method for either.

Brainlabs ran 23 tests across 16 mature advertisers over nine months, written up in Search Engine Land in March 2026. Campaigns using all three features had a 40% higher uplift in test success rates than those using search term matching alone. The same write-up reports that only 46% of AI Max queries were new to the account. The other 54% had been captured elsewhere before.

Smarter Ecommerce looked at more than 250 retail Search campaigns from over 600 accounts. Its March 2026 guide reports a median 13% more conversion value, a median 16% higher cost per acquisition, and a median ROAS difference of exactly 0%, with campaigns ranging from 42% above baseline to 35% below. Search Engine Journal's summary adds that only 22% of campaigns landed close to their original ROAS target.

Brad Geddes tested text customization at three companies and published the results in July 2026. The ecommerce account looked like a win until the account view showed the campaign taking volume from others, and revenue fell. The B2B account drew consumer searchers, saw its conversion rate drop, and stopped after three weeks. The B2C lead generation account gained in neglected long-tail campaigns.

Both sides can be right. Google's figures count platform conversions at a similar cost, on average, outside retail. The independent sets show a flat median with a wide spread, and they measure something Google's claim does not: whether the conversions were new. An average cannot tell you which side of the spread your account is on. Only your own data can.

Why the AI Max label is not proof of new conversions

The Keywords report now ends with two summary rows, Total: AI Max expanded matches and Total: AI Max landing page matches. The search terms report has an AI Max match type. Use them, with three cautions.

First, the label marks how a query was matched, not whether you would have won it anyway. The Brainlabs figure above is the warning.

Second, the rows can undercount. Google's help page notes that filtering the search terms report by the AI Max match type may show lower numbers. Adalysis reports that when no broad keyword exists, Google assigns AI Max data to phrase or exact match in most cases.

Third, platform conversions are not customers. RazorRank pooled its law firm accounts and found conversions up 28% after AI Max while signed cases fell from 20 to 12. That is a small count, so treat it as a prompt to check your own pipeline. The same gap between a form fill and a real lead drives the sum in our piece on gating a guide.

So read two things: the AI Max rows, and the campaign total before and after. The difference between a platform label and added outcomes is the same problem we covered for Meta's attribution settings.

How much data before you decide?

Conversions arrive as random counts. A count of n carries a 95% range of roughly plus or minus 2 divided by the square root of n. These are our calculations, not Google's.

Conversions on AI Max rowsApproximate 95% range on their CPA
10Plus or minus 63%
30Plus or minus 37%
50Plus or minus 28%
100Plus or minus 20%

A workable rule: read the AI Max rows once they have spent 30 times your target CPA. On target, that buys 30 conversions. If you see 23 or fewer, the cost is about 30% above target or worse, and you trim. By our calculation this rule fires about 11.5% of the time when AI Max is truly on target, and catches a true cost 50% above target about 79% of the time. It is the same sample logic we used to size creative tests.

Geddes's field rule points the same way. He describes AI Max as highly erratic under 30 conversions a month and good more often than not above 100. That is experience, not a dataset. A campaign that needs a quarter to reach the read point is too small to judge, and we would switch search term matching off there.

Which setting stays on: a decision table

SettingAfter migrationKeep it on whenTurn it off when
Search term matchingOn for both groupsConversion bidding is in use, the budget is not limiting, and the read rule passes at campaign level tooBidding is manual CPC, where Google says it will not work. The campaign is limited by budget. The read rule fails. Lead quality drops
Text customizationOn for the former automatically created assets group. Off for the broad match groupAd groups are long-tail and under-resourced, and someone reviews the assetsCopy must prequalify buyers, carry fixed legal wording or stay pinned
Final URL expansionReported off for both groupsURL exclusions cover blog, careers and policy pages, and tracking templates are testedYou have not set exclusions. Turning text customization off disables it automatically

Google's help centre is blunt about one condition: "AI Max won't be effective if campaigns are limited by budget."

One argument for keeping everything on has weakened. Ads in AI Mode used to need AI Max, Performance Max or broad match with Smart Bidding. On September 4, 2026 Search Engine Roundtable reported that Google's Ads Liaison, Ginny Marvin, had confirmed a small experiment in which exact and phrase keywords can serve there when intent is explicit and direct. It is a test, and Google has not said it will last. If AI answers matter to your category, the waitlist is where we will share how we measure brand presence in them.

A seven-step method

  1. Find the migrated campaigns. Open the AI Max section of campaign settings and note which of the three features is on.
  2. Check the preconditions. Conversion-based bidding, no budget limit, and ideally 30 or more conversions a month.
  3. Fence it. Negative keywords still apply. Add brand exclusions. In the former broad match group, Google's help page says broad keywords are no longer treated as exact for prioritisation after the move, so add exact match copies of your most valuable keywords. Where text customization is on, set text guidelines, which allow up to 25 term exclusions and 40 messaging restrictions per campaign, and remove any generated asset that is wrong.
  4. Fix the rule before you look. 30 times target CPA in spend on the AI Max rows. Trim at 23 or fewer conversions.
  5. Read the rows. Use the two summary rows in the Keywords report.
  6. Read the campaign. Compare total conversions and spend with an equal period before migration. Divide the extra spend by the extra conversions. Then check lead or order quality downstream.
  7. Decide per ad group. Search term matching can be switched off for one ad group and left on for another.

A cleaner test exists. Google's AI Max experiment splits traffic inside one campaign between AI Max off and on. It cannot be created while text customization is enabled, or with a portfolio bid strategy or shared budget, so many migrated campaigns do not qualify as they stand.

A worked example with invented numbers

This company is hypothetical. It sells B2B software and had the broad match setting on one campaign. In the four weeks before migration it spent $12,000 for 80 demo requests, a CPA of $150, its target.

In the four weeks after, the campaign spent $13,500 for 86 requests. The AI Max rows show $4,500 in spend, which is 30 times target, and 22 conversions. Their CPA is about $205, or 36% above target. The rule says trim.

The campaign view is worse. Keyword rows fell from 80 conversions to 64. The 22 AI Max conversions replaced 16 of them, so the net gain is 6 requests for $1,500 more spend. That is $250 per extra request, 67% above target. Only 6 of the 22, about 27%, were new.

The company switches search term matching off in the ad groups carrying most of that spend and leaves the rest running. A before-and-after comparison is not an experiment, so it repeats the read after the next 30 times target.

Where this stops

Reporting will improve. On September 23, 2026 Google announced a single report linking search term, assets shown and landing page, with availability promised later this year. Until then the two summary rows and the campaign total are the tools.

Microsoft has followed. Its own AI Max for Search became available to all advertisers in August 2026 and is on by default for new campaigns. Search Engine Journal reports it was switched on automatically for campaigns already using predictive matching or autogenerated text assets. The same three decisions apply there.

Questions

Which campaigns did Google move to AI Max in September 2026?

Search campaigns that used the campaign-level broad match setting or automatically created assets. Google stopped new use of both on August 3, 2026 and migrated existing campaigns in place between September 1 and 30. Dynamic Search Ads are separate. Their automatic migration was pushed back to February 2027.

Can I turn AI Max off after the automatic upgrade?

Yes. One campaign setting switches all of AI Max off, and each of the three features has its own control. Search term matching can also be switched off per ad group. Google's help centre says turning text customization off also turns final URL expansion off. The old legacy settings cannot be recreated.

How long should I wait before judging AI Max?

Size the read by spend, not by days. Wait until the AI Max rows in the Keywords report have spent about 30 times your target CPA. Before that, the cost per conversion carries a range wider than 37% either way, by our calculation, and a decision is mostly a guess.

Do I need AI Max for my ads to show in Google's AI Mode?

Less than before. On September 4, 2026 Google's Ads Liaison confirmed a small experiment in which exact and phrase match keywords can serve text ads in AI Mode when intent is explicit. Before that, AI Max, Performance Max or broad match with Smart Bidding was needed. Google has not said the test is permanent.

Sources

  1. Google, Google's Dynamic Search Ads are upgrading to AI Max blog.google
  2. Google, Unlock next-level performance with AI Max for Search campaigns blog.google
  3. Google Ads Help, How AI Max for Search campaigns works support.google.com
  4. Google Ads Help, Set up AI Max for Search campaigns support.google.com
  5. Google Ads Help, Frequently asked questions about AI Max for Search campaigns support.google.com
  6. Google Ads Help, About reporting in AI Max for Search campaigns support.google.com
  7. Google Ads Help, About AI Max experiments support.google.com
  8. Google Ads Help, About the broad match keywords campaign setting support.google.com
  9. Google Ads Help, Text customization frequently asked questions support.google.com
  10. Google Ads Help, Use text guidelines with Performance Max and search campaigns (beta) support.google.com
  11. Google, New AI Max tools can help you scale Search campaigns business.google.com
  12. Google, We're bringing AI Brief to more languages and adding a new AI Max reporting feature blog.google
  13. Microsoft Advertising, AI Max for Search is now available about.ads.microsoft.com
  14. Search Engine Land, Google sets AI Max migration timeline for Search campaigns searchengineland.com
  15. PPC Land, Google Ads broad match campaigns face AI Max auto-upgrade on September 1 ppc.land
  16. Choice Marketing, AI Max by Default, The Verified Dossier on Google's September Auto-Upgrade choice.marketing
  17. Smarter Ecommerce, The Ultimate Guide to AI Max for Google Search smarter-ecommerce.com
  18. Search Engine Journal, What SMEC's Data Reveals About AI Max Performance searchenginejournal.com
  19. Search Engine Land, What 23 tests reveal about Google AI Max performance searchengineland.com
  20. Search Engine Land, Putting Google Ads AI Max's automated ad copy to the test searchengineland.com
  21. Search Engine Land, Is your account ready for Google AI Max? A pre-test checklist searchengineland.com
  22. Adalysis, How to evaluate your AI Max performance adalysis.com
  23. RazorRank, AI Max for Search, More Conversions, Fewer Cases razorrank.com
  24. Lunio, AI Max and Invalid Traffic, The Findings from 414 Million Retail Ad Clicks lunio.ai
  25. Search Engine Roundtable, Google Ads Testing Serving Restrictive Match Types In AI Mode seroundtable.com
  26. Search Engine Land, Google tests traditional Search campaigns in AI Mode searchengineland.com
  27. Search Engine Land, Google expands AI Brief and adds deeper AI Max reporting searchengineland.com
  28. Search Engine Journal, Microsoft Advertising Rolls Out AI Max Globally searchenginejournal.com
  29. Brandrocket, Google AI Max Auto-Upgrade, What Changed in Your Account brandrocket.net
  30. Platform81, Google Just Upgraded Your Search Campaigns to AI Max, What to Check This Week platform81.com
E
Esh

Founder, Ethos Engine. Builds the Digital Authority Score and runs the placement marketplace. esh@ethosengine.io

NEXT STEP

See where your brand stands.

Your Digital Authority Score, who AI recommends for your target questions, and the placements that close each gap. Your first report is free.

Get your free report →
MOREINSIGHTS