Google Is Migrating Your Broad Match Campaigns to AI Max This Month: What to Check Before Yours Flips
By Search Solutions LLC • September 2026 • 9 min read
On September 1, Google started rewriting the settings inside Search campaigns that nobody asked it to touch. If your account uses the campaign-level broad match setting, or standalone Automatically Created Assets, Google is upgrading those campaigns to AI Max — automatically, in place, and without an opt-in. The rollout runs progressively through September 30, which means a meaningful share of accounts reading this haven’t flipped yet. Yours might change tomorrow. It might change next Tuesday.
There is no opt-out. The only way to avoid the migration is to turn off the underlying features entirely, which for most advertisers is a worse outcome than the migration itself. Google closed the door on new legacy setups back on August 3 — you can no longer create campaign-level broad match configurations or legacy Automatically Created Assets in the interface, in Google Ads Editor, or through any version of the API.
So the question isn’t whether to accept this. It’s what to do in the two weeks on either side of the flip so that a settings change made by an algorithm doesn’t quietly rewrite your cost per lead. That’s a very different question than the one most of the coverage is answering — and the independent performance data on AI Max is a lot more interesting than the headline number Google is promoting.
| Sept 1–30 the window in which Google auto-upgrades campaign-level broad match and Automatically Created Assets to AI Max | +16% median cost-per-acquisition increase alongside a +13% median conversion value lift, across 250+ campaigns in an independent 2026 study | 22% share of those campaigns that landed anywhere near their original ROAS target — the other 78% swung hard in one direction or the other |
What Actually Changes Inside Your Account
AI Max isn’t a new campaign type. It’s a bundle of three settings that sit on top of a standard Search campaign: search term matching, text customization, and final URL expansion. Search term matching is the significant one — it uses broad match and keywordless technology to find queries your keyword list would never have matched, learning from your existing keywords, assets, and landing pages.
What matters for the migration is which of those three switches Google flips on for you, and the answer depends on which legacy feature triggered your upgrade. Google’s developer documentation spells out the default migrated configurations, and they are not identical:
If campaign-level broad match triggered it — search term matching comes on. Text customization and final URL expansion stay off. Your existing brand inclusions and exclusions are preserved and carried into the new setup.
If Automatically Created Assets triggered it — search term matching comes on and text customization comes on by default. Final URL expansion stays off. That second switch means Google can generate new headlines and descriptions from your landing pages and keywords.
Dynamic Search Ads are not part of this wave — active legacy DSA campaigns are excluded from September and keep serving normally. Their automigration is scheduled for February 2027, with in-account warning banners appearing this month and reminder notifications in mid-January.
Google’s stated goal is continuity — migrating in place with equivalent settings to minimize performance volatility. That’s a reasonable design choice, and it’s better than the alternative. But “minimize volatility” is not “no change.” Search term matching is, by definition, a change in what queries you buy. Search Engine Land’s timeline of the migration also flags a detail worth noting if you run scripts or third-party tools: any Google Ads API version released after September 1 removes the legacy entities entirely, and older versions sunset around September 2027.
“An in-place migration with equivalent settings is still a migration. The switch that comes on by default is the one that changes which searches you pay for.”
The Honest Limitation: The Benchmarks Don’t Say What You Think They Say
Google’s headline claim for AI Max is that advertisers who activate it typically see 14% more conversions or conversion value at a similar CPA or ROAS, rising to 27% for campaigns still running mostly exact and phrase keywords. Those numbers are real. They are also carrying more qualifications than the headline suggests, and the footnote does the heavy lifting: the figure comes from Google internal data, restricted to non-Retail advertisers, and only counts campaigns where more than 70% of conversions came from exact or phrase match keywords.
That is a narrow population. If you’re an e-commerce advertiser, or you already lean on broad match, or your conversion mix doesn’t look like that, the 14% was never measured on an account like yours. It isn’t a wrong number. It’s an answer to a question you may not be asking.
The volume comes at a higher unit cost. Across 250+ campaigns, the median conversion value lift was +13% — strikingly close to Google’s figure — but the median CPA rose 16% at the same time. More conversions, each one more expensive.
The median hides an enormous spread. ROAS outcomes ranged from 42% above baseline to 35% below it. Only 22% of campaigns landed close to their original target. A median of “roughly neutral” describes almost none of the individual accounts inside it.
Expansion overlaps with what you’re already running. Roughly half the accounts studied ran AI Max, Dynamic Search Ads, and Performance Max simultaneously — three systems all designed to reach beyond your keyword list, competing for the same queries and splitting conversion data across campaigns.
The two datasets aren’t directly comparable. Google’s benchmark excludes Retail; the independent study is heavily e-commerce. Both can be accurate and still not describe your lead-gen account. Treat neither as a forecast.
One more finding is worth sitting with, because it contradicts the mental model most advertisers have. Analyzing a million AI Max impressions, that same research found expansion came overwhelmingly from exact match keywords — roughly 80% — with under half a percent originating from broad match. Search Engine Journal’s breakdown of the SMEC study frames the practical implication well: AI Max takes tightly defined keywords and widens the set of queries it considers relevant. It is not simply broad match with better branding.
Where AI Max Genuinely Helps
None of the above is an argument that AI Max is bad. It’s an argument that it’s a volume expansion layer, and volume expansion is genuinely valuable when your problem is reach and you have the margin to absorb a higher cost per acquisition. Three situations where it earns its keep:
You’ve plateaued on impression share. If your core keyword set is fully covered, your budget is capped by available volume rather than by efficiency, and you’ve been asking where the next 20% of leads comes from — this is a real answer. Net-new queries you were never eligible for are the entire point, and Google’s own case studies show meaningful conversion volume arriving specifically from queries that didn’t exist in the account before.
Your keyword list has gone stale. Most accounts we audit have a keyword set built two or three years ago and lightly maintained since. Language changes. Products change. The way people phrase a problem changes. Search term matching surfaces the vocabulary your prospects actually use right now, and even if you decide the expansion isn’t worth the CPA, the search terms report it produces is a research asset worth having.
You have strong conversion signal and honest conversion actions. AI Max optimizes toward whatever you told it to value. Accounts with clean conversion tracking, deduplicated lead events, and offline outcomes fed back into the platform give the system something real to chase. That’s exactly where the wide ROAS spread starts collapsing toward the good end — and it’s the single biggest lever most advertisers still haven’t pulled.
Your September Checklist: What to Do Around the Flip
The migration is a settings change you didn’t choose. The response is a measurement discipline you do choose. Here’s what a well-run Google Ads program should be doing in the next four weeks:
Write down your baseline today. Before your campaigns flip, export the last 30 and 90 days of cost, conversions, cost per conversion, and conversion rate at the campaign level. Note the exact migration date when it lands — Google records it on the campaign, so you can query it rather than guess. Without a clean pre-migration baseline, every performance conversation for the next quarter becomes an argument about vibes.
Read your search terms report weekly, not monthly. Expansion means new queries. Some will be excellent and some will be job seekers, students, competitors, and people looking for a free version of what you sell. A weekly review with disciplined negative keyword additions is the difference between AI Max finding you customers and AI Max spending your budget on curiosity.
Audit your overlap before you audit your results. If you’re running AI Max, Dynamic Search Ads, and Performance Max at once, you have three systems bidding on the same expanded query space. Decide deliberately which one owns discovery. Cannibalization between your own campaigns will look exactly like an AI Max performance problem, and you’ll fix the wrong thing.
Judge it on qualified leads, not conversions. This is the one that separates accounts that win from accounts that get louder. A form fill is not a customer. If your CRM outcomes aren’t flowing back into Google, the algorithm is optimizing toward volume you may not want. Fixing that feedback loop is the highest-leverage work in most lead generation programs right now, migration or no migration.
“Google decides what your campaigns do. You still decide what counts as a win. The accounts that come through this migration ahead are the ones that defined that clearly before the switch flipped.”
The Honest Bottom Line
Between September 1 and September 30, campaign-level broad match and Automatically Created Assets stop existing as separate features and become AI Max settings. You can’t opt out, only opt down. Google says the migration is designed for continuity, and there’s no reason to doubt that intent — but the feature you’re being migrated into produced a +13% median conversion value lift and a +16% median CPA increase in the largest independent study available, with only 22% of campaigns landing anywhere near their prior ROAS target.
That spread is the real story. It says outcomes depend far more on account structure, conversion signal quality, and campaign overlap than on the feature itself. Which is another way of saying the migration will amplify whatever your account already is. Well-structured programs with clean measurement will likely find incremental volume worth paying a bit more for. Messy accounts with duplicated campaigns and conversion actions that count everything will find the expansion expensive and the reporting harder to read than before.
You have a narrow window this month to set a baseline, tighten your conversion definitions, and decide which campaign owns query discovery. Do that work now and AI Max becomes a reach lever you control. Skip it and you’ll spend the fourth quarter trying to explain a cost-per-lead change you can no longer trace.
Want to Know What the AI Max Migration Is Doing to Your Cost Per Lead?
We’ll baseline your Google Ads account before the switch flips, audit your campaign overlap, and tell you plainly what’s working and what isn’t. No contracts. No hidden costs. Complete transparency.