Google's forced AI Max migration starts in September 2026. That's not a roadmap slide, it's the window Google committed to at Marketing Live in May. Dynamic Search Ads, Automatically Created Assets, and campaign-level broad match are all being absorbed into AI Max. Gemini takes over query matching, text generation, and landing page selection. You'll get a few new control knobs back, but most of what advertisers used to manage by hand is going under the bonnet.

Which means the next three months matter, and the question worth asking is whether your Google Ads account is actually AI ready. If you're feeding AI the right signals, the migration is genuinely upside; Google's own data suggests advertisers with strong first-party data lift incremental ROAS around 11%. If you're feeding AI bad signals, AI Max will scale those bad signals across query selection, ad copy, *and* landing page choice. It's a force multiplier in both directions.

This post is the diagnostic I run on every new client. It's a six-layer framework that scores an account out of 100, with a Foundation gate that caps the score at 40 if certain things are broken. The version below is the lean explainer; the full rubric I score on lives in our delivery process.

What "AI Ready" Actually Means

The phrase gets thrown around like it means "modernised". It doesn't. AI ready means the account can hand decisions to Smart Bidding (and soon AI Max) without amplifying mistakes. Three things have to be true:

The conversion data Google sees is what actually matters to your business.
Google can tell good leads from bad ones, not just count form fills.
The structural inputs AI relies on - landing pages, audience signals, creative assets - are actually populated.

Most accounts I audit fail at least one of those, and many fail all three. The rest of this post walks through the six layers I score on, with rough indicators for each.

Layer 1: Foundation (40 of 100 points)

This is the heaviest layer for a reason. Broken foundations cap everything above. The full layer has nine criteria including Enhanced Conversions, conversion values, and managed auto-created conversions, but four of those criteria act as a hard gate:

Google Tag installed sitewide. Not "installed on the thank-you page". Sitewide, firing on every page, with no JavaScript errors.
Primary conversion is a meaningful lead event - form submission, qualified call, booking confirmation. Not a pageview. Not a button click. Not "scroll past hero".
Offline Conversion Import is flowing in the last 7 days. This is the one most accounts miss. If your CRM knows which leads turned into customers and Google doesn't, you're optimising the wrong thing.
A lead-quality conversion is defined - SQL, qualified, won, or weighted by value.

If any of those is at zero, your total score caps at 40. Doesn't matter how good your structure or creative is. AI optimises to whatever you tell it the primary is. If the primary is a pageview, Smart Bidding will buy you pageviews. It's that direct.

I scored a UK lead-gen account in May this year. They had a tidy structure, decent audiences, polished ads. Total score: 54/100. Gated by missing OCI from their CRM. Their 18-month sales cycle meant Google was bidding on form fills with no signal at all about whether those fills became customers. The fix wasn't a campaign rebuild. It was a pipeline build.

Layer 2: Wiring (20 points)

This is the audience signal layer - the data you give Google to find more people like your good customers.

Customer Match list active and applied to acquisition campaigns, not just remarketing. Refreshed within the last 90 days. Not applicable if you have under 100 monthly conversions or fewer than 1,000 matched customers.
GA4 audiences imported and used as signals. Not applicable if you have under 30 monthly conversions - the signal's too thin to be useful.
In-market and affinity audiences applied where relevant to your industry.
Custom intent or custom segments for accounts where competitor or category targeting makes sense.

The point of this layer is that Smart Bidding doesn't just optimise on conversion data - it uses audience signals to find *new* people who look like your converters. Without those signals, AI is fishing in a smaller pond.

Layer 3: Structure (12 points)

This used to be a 20-point layer in the previous version of the framework. AI Max collapsing match-type and ad-group boundaries cost it eight points. What's still load-bearing:

Brand and non-brand separated into different campaigns with different bidding strategies. Mixing them is the fastest way to make Smart Bidding decisions look better than they are.
Bidding strategy fits conversion volume - at least 30 conversions in the last 30 days for tCPA, 50 for tROAS. Below that, Smart Bidding doesn't have enough data to train.
Negatives organised in shared lists, not chaos. This matters *more* under AI Max, not less. Negatives are one of the few query-routing controls advertisers will keep.
Geo and schedule signals appropriate - not blanket-applied across campaigns that have nothing in common.

Themed ad groups still help reporting and negative discipline, but they've stopped being a performance lever in their own right. Don't fund a restructure project on tighter ad groups when Foundation is broken. The order of operations matters here, and structure is third on the list, not first.

Layer 4: AI Max Readiness (8 points)

This is the new layer in the framework, added after Marketing Live 2026. It tests whether you've got the controls AI Max will actually give you and the inputs it depends on.

Landing page coverage - your site has dedicated pages for major service or product variants, not one generic catch-all. AI Max picks the landing page from your site. If there's only one page, every query lands on the same place no matter the intent.
URL controls configured - page feeds with custom labels, URL inclusions at ad group level, URL exclusions at campaign level. These are the levers AI Max gives you to keep traffic on appropriate pages.
Brand safety controls - an account-level negative keyword list plus documented brand text guidelines for AI-generated copy.
DSA pre-migration baseline captured if you've got active Dynamic Search Ads campaigns. You'll want the before/after data when migration happens.

The landing page point is the big one. The advertisers who'll do best out of AI Max are the ones whose websites already help Google make good routing decisions. If your site is a single landing page with a contact form, AI Max can't do much for you no matter how much you optimise the account.

Layer 5: Creative (10 points)

Creative used to be about per-ad strength. Under Asset Studio and AI Max, it's about library depth.

RSA strength Good or Excellent on more than 80% of ads. Still tested, lower weight than it used to be.
Sitelinks and callouts - at least four of each per campaign.
Structured snippets where applicable.
Lead form extensions on relevant campaigns. Not applicable if you use a third-party booking widget or sit in a regulated vertical.
Asset library depth - text, image, and video assets across the formats Google now expects. This is where most accounts are thin.

Google rotates and remixes assets now. The differentiator isn't whether one ad scores Excellent. It's whether you've given Asset Studio enough to work with.

Layer 6: Plumbing (10 points)

The boring layer that gates everything else. Usually binary - either it works or it doesn't.

Landing page matches ad copy intent - no bait-and-switch between ad and page.
Core Web Vitals pass on the landing pages your ads point to.
Form quality - appropriate fields, mobile UX that doesn't fall over.
Base GTM tag health - container loaded sitewide, no JavaScript errors.
Tracking completeness - every form path you know about fires a Google Ads conversion. No documented broken tags.

If your conversion tracking has decayed, this is where it shows up. Tracking decay is also why so many accounts score worse on Foundation than they should - the data they think Google sees isn't the data Google actually sees.

How to Score Your Own Account

Rough bands once you've added it up:

0-40: Not Ready. Foundation gate is almost certainly the issue. Fix tracking, conversion truth, OCI flow, and lead-quality signal *before* touching anything else.
41-70: Approaching Ready. Foundation is intact, but you're missing audience signals, AI Max prep work, or creative depth. The September migration window is enough time to close the gap if you start now.
71-100: Ready. AI Max migration should be net upside. Use the window to capture pre-migration baselines on any DSA campaigns and document brand text guidelines.

Most accounts I see in audits sit somewhere between 35 and 60. That's not a failure - it's the gap between how Google Ads worked five years ago and what AI Max is about to ask of you.

What to Actually Do Next

My take: most of the "get AI ready" advice doing the rounds is repackaged automation evangelism. Google selling AI Max as upside is only true if your account is already in the top quartile on signal quality. Everyone else is being sold a wider distribution of the problems they already have. The fix isn't more automation, it's better inputs.

If you're going to spend the next three months on one thing, it should be the Foundation layer. Specifically, get Offline Conversion Import flowing from your CRM. Most lead-gen businesses I work with had no OCI when I started; it's the single highest-impact move for reducing wasted ad spend and the one that unlocks every layer above it.

If Foundation is intact, spend the time on landing page coverage. AI Max performance is going to be capped by the worst page on your site, and that's not a problem the platform can solve for you.

You can read Google's own Marketing Live 2026 recap for the official line. The practical version is what's above.

Key Takeaways

Forced AI Max migration starts September 2026. The next three months are the prep window.
AI is only as good as the signal you feed it. Bad signal scales the same way good signal does.
The Foundation gate is non-negotiable. If conversion tracking, lead-quality signal, or OCI is broken, score caps at 40.
Audience signals (Customer Match, GA4 audiences) are how Google finds new people like your good customers. Skip this and AI fishes in a smaller pond.
Landing page coverage is the highest-impact AI Max prep work. One generic page caps your performance no matter what you do with bids.

If you want to know where your account actually sits on this framework, that's the kind of audit I run for clients. The output is a scored report plus a sequenced migration plan - not a slide deck, an actual punch list. Get in touch if you'd like one.

Getting Your Google Ads Account AI Ready: A Diagnostic Framework