Google vs. Microsoft AI Max: What Operators Must Know

TL;DR: Google and Microsoft AI Max share three core features — search term matching, text customization, and final URL expansion — but differ on transparency, brand control limits, and how settings are structured. Operators who treat these platforms as interchangeable will waste budget and lose control of messaging. This breakdown shows what actually matters for accounts running real spend.

What AI Max Actually Is (and What It Is Not)

AI Max is not a new campaign type. It is a set of optional settings that sit inside standard Search campaigns on both Google and Microsoft. That distinction matters. Unlike Performance Max or Demand Gen, AI Max does not replace your campaign structure — it layers on top of it. Advertisers who misread it as a separate campaign type will configure it wrong from the start.

The three settings are designed to work together. Search term matching expands your reach beyond your static keyword list by pulling in intent and contextual signals from your ads, landing pages, and existing keywords. Text customization uses your existing assets and site content to generate messaging variations and select the best combinations at auction. Final URL expansion routes users to the page on your site that best matches their query, rather than a fixed landing page.

When all three run together, they create a closed loop: a broader query pool, ads that adapt to those queries in real time, and landing pages that match what was promised in the ad. Operators running paid search campaigns at scale will recognize this as the logical end state of smart bidding — just extended to creative and routing.

What Is Identical Across Both Platforms

The shared mechanics go deeper than the three headline features. Both Google and Microsoft require conversion-based bidding to enable search term matching. That is not optional. The platforms need conversion signals to decide which expanded queries are worth entering. The recommended floor is 15 to 30 conversions in a 30-day window before enabling this feature. Below that, the system lacks the data to make defensible decisions.

If your account is below that threshold, two paths exist. First, wait until volume catches up. Second, use micro-conversions with assigned values at each funnel stage. For example, a financial product campaign might assign $10 to a started application, $20 to a mid-funnel completion, $50 to a finished application, and actual deal value (via offline conversion upload) to a funded account. This gives the bidding algorithm something real to optimize against without waiting for closed conversions to accumulate.

Both platforms also support brand controls — inclusions, exclusions, term exclusions, and message constraints. The mechanics are the same in concept. Where they differ is in the limits, covered below.

Both support A/B experiments before full rollout. Start with your strongest campaign — one with stable performance and enough traffic volume. Run a 50/50 split and give the test time to learn. Ending a test after one week because CPA ticked up is the most common way operators misread AI Max experiments. Auditing your account structure before launching any AI Max test is worth the time — accounts with fragmented ad groups or inconsistent conversion tracking will produce noisy experiment results.

Where Google and Microsoft Diverge

The differences are real and operationally meaningful. Here is where they split:

Brand control limits: Google allows 10 brand lists per campaign with up to 5,000 brands per list. Microsoft allows 20 brand lists per campaign but caps each list at 100 brands. For operators managing large exclusion lists — common in regulated iGaming acquisition or financial services — that 100-brand cap on Microsoft is a hard constraint that requires a different list architecture.

Search term transparency: Microsoft provides full search term reporting for every query that results in a click, across AI Max, PMax, and standard Search. Google hides some search terms for privacy reasons, which limits your ability to audit whether expanded queries are relevant. Google partially compensates by supporting more granular negative keyword close-variant matching. For operators who rely on search term data to refine targeting — standard practice in law firm lead generation where irrelevant clicks carry high CPCs — Microsoft’s full reporting is a concrete advantage.

Setting architecture: Microsoft keeps all AI Max settings at the campaign level. Google pushes some decisions down to the ad group level, including URL inclusions, brand inclusions, and locations of interest. This means Google requires more active management per ad group, while Microsoft offers a simpler campaign-level toggle structure. For large accounts with many ad groups, Google’s approach creates more control surface but also more maintenance overhead.

Matching signals: Google pulls from YouTube viewing behavior, in-market audiences, Customer Match, demographics, and conversion data. Microsoft pulls from LinkedIn profile data, impression-based remarketing, in-market audiences, and conversion data. The LinkedIn signal is meaningful for B2B operators and any vertical where professional role or industry matters for qualification. Audience-level targeting on Microsoft can be sharper for business-oriented offers precisely because of that LinkedIn data layer.

Disclaimers: Microsoft supports disclaimers that do not consume ad real estate and work natively with AI Max. Google is still piloting disclaimers as of this writing, with the current version taking up description line two. For regulated industries where disclaimers are legally required, Microsoft’s current implementation is less costly in terms of available ad space.

How to Structure an AI Max Test Without Wrecking Performance

The experiment guidance from both platforms converges on the same principles: start with a campaign that already performs, split traffic 50/50, and give the system time to accumulate signal before reading results. The typical mistake is pulling the plug in week two because one metric looked wrong.

For operators who want to test conservatively, Final URL Expansion plus Text Customization is a reasonable starting point. Disabling search term matching keeps the query pool tighter while still allowing the system to optimize creative and routing. An ecommerce or lead gen operator with consistent margin across offers is a natural candidate for this configuration.

If you are running search term matching, get your conversion tracking correct first. Broken or misfiring conversion tags will corrupt the bidding model and produce results that look like AI Max failures but are actually data quality failures. AI-powered lead qualification downstream of your campaigns also depends on clean conversion definitions — if the platform’s definition of a conversion does not match what your sales team counts as a lead, every optimization decision the algorithm makes will be pointed at the wrong target.

What This Means for High-CAC Verticals

Operators in forex, crypto, iGaming, and legal work with high cost-per-acquisition and tight compliance constraints. AI Max introduces real efficiency gains but also real risks for these verticals specifically.

For forex acquisition campaigns, the expanded search term matching can surface intent signals that keyword lists miss — particularly around conversational queries like “how do I trade oil CFDs without a broker” that no static keyword would catch. The risk is that expanded matching also pulls in unqualified traffic if brand controls and negative lists are not properly configured before launch.

For crypto and web3 lead generation, Microsoft’s full search term transparency is valuable. In a vertical where regulatory language shifts fast and one wrong ad placement can trigger a policy violation, being able to audit every query that drove a click is not a nice-to-have — it is a compliance requirement.

Legal operators should pay close attention to the disclaimer handling difference. If your campaigns require legal disclaimers on every ad — common in mass tort and personal injury — Microsoft’s current implementation preserves more of your description line real estate. Running those campaigns on Google with disclaimer pilots enabled could reduce effective copy by one full description line.

The conversion threshold requirement (15 to 30 per 30 days) also hits high-CAC operators harder. A personal injury firm with a 3-week intake-to-signed-client cycle may struggle to accumulate enough conversion signal in the window the platform expects. Micro-conversion mapping — from form fill to consult booked to retainer signed — is not optional in these accounts; it is the only way to give AI Max enough signal to function.

Run a structured performance audit before enabling AI Max across any high-CAC account. Know your conversion volume by campaign, verify your tracking stack, and map out your brand exclusion lists before you flip the switch.

Originally reported by Search Engine Land, September 2026.