Google has been iterating on AI Max for Search controls through 2026, adding capabilities that address the specific objections B2B SaaS advertisers raised when the feature first launched. The three most material additions — brand keyword exclusions, audience signal inputs, and URL-level performance reporting — change the risk calculus for enabling AI Max in B2B accounts where brand protection and signal quality are non-negotiable.
This is a different question from whether AI Max performs well in aggregate. The 2026 performance data on AI Max for B2B SaaS covers that in detail — the short version is that it works in accounts with strong offline conversion signal and controlled URL structures, and underperforms in accounts without those conditions. This post focuses specifically on the new configuration levers available and how to use each one in a B2B SaaS account.
Brand keyword exclusions: protecting your branded campaigns
The single most common AI Max complaint from B2B SaaS advertisers is brand cannibalization: AI Max query expansion finding its way into branded search territory and serving non-branded CPCs on queries that your branded campaign should be handling at a fraction of the cost. Without controls, AI Max treats your brand name as a relevant expansion target — which is technically correct from an intent standpoint but expensive from an economics standpoint.
Brand keyword exclusions in AI Max are set at the campaign level under the AI Max settings section of the campaign. You specify brand names (including common misspellings and abbreviations) that the AI should not expand into. This is distinct from standard negative keywords, which block specific search term strings — brand exclusions block the entire brand-intent space, including novel phrasings the algorithm might generate that you have not explicitly listed. For B2B SaaS, set exclusions for your company name, product name, domain name, and any commonly used abbreviations. If you run competitor campaigns, exclude competitor brand names from your non-competitor AI Max campaigns to prevent cross-contamination.
After setting brand exclusions, pull your search term report for AI Max campaigns weekly for the first month and verify that branded queries are not appearing. If they are, check that the exclusion list covers all spelling variations and that you have not inadvertently set a campaign-level brand restriction that overrides the exclusion.
Audience signal inputs: steering AI Max toward your ICP
AI Max audience signals are first-party audiences you provide as training input — remarketing lists, CRM customer matches, and similar segments built from your best converters. They do not restrict who AI Max can reach; they bias the algorithm toward users who resemble the people in those lists. The practical effect in B2B SaaS is that query expansion and URL selection skew toward higher-intent traffic segments rather than the broadest possible audience.
The highest-value signal for B2B SaaS is a closed-won customer list uploaded as a customer match audience. If you have at least 1,000 matched email addresses, Google can build a similar audience from that list and use it as a training signal. The second most valuable is a list of SQL-stage leads — people who filled out a form and were subsequently qualified, not everyone who clicked. If you only have raw form fill remarketing lists, the signal quality is lower and you should invest in building the SQL list before leaning heavily on audience inputs.
For accounts where customer match is not possible (typically under $50K monthly spend or accounts without sufficient historical customer data), use your site visitor remarketing audiences segmented by page — visitors to your pricing page or demo page are higher quality signals than all visitors. Combined with the brand keyword exclusions, audience signals effectively give AI Max a shaped learning environment rather than a blank slate.
URL-level reporting: auditing where AI Max sends traffic
URL expansion is the AI Max feature with the highest risk for B2B SaaS accounts. When enabled without restrictions, AI Max can send paid traffic to any page on your domain it judges relevant — blog posts, documentation, support articles — that were not designed as paid acquisition landing pages and typically convert at a fraction of the rate of purpose-built paid pages. URL-level reporting is the tool that makes URL expansion manageable rather than opaque.
Find URL-level reporting under the AI Max insights section within the campaign view. The report shows impressions, clicks, conversions, and conversion rate broken down by the destination URL that was served. Run this report covering the last 30 days on any AI Max campaign that has URL expansion enabled. Sort by clicks, descending. For every URL receiving more than 50 clicks, check the conversion rate against your account average for dedicated landing pages.
The threshold for exclusion is not zero conversions — it is relative underperformance. A blog post that converts at 0.8% when your landing pages convert at 12% is a drag on overall campaign performance even if it is generating some conversions in absolute terms. Add underperforming URLs to your URL exclusion list within the AI Max settings. Repeat this audit monthly until your URL distribution stabilizes on high-converting pages. At that point, you will know that AI Max is operating within a controlled URL set and the URL expansion is adding volume, not diluting conversion rate.
Text guidelines and messaging controls
AI Max text guidelines allow you to set constraints on what the creative generation feature can say in auto-generated ad copy. The current limits are up to 25 term exclusions (words or short phrases the AI cannot use in generated text) and up to 40 messaging restrictions (broader guardrails on tone, claims, or positioning). Both are set at the campaign level.
For B2B SaaS, text guidelines are most useful for preventing three categories of problematic generated copy: inflated performance claims the AI might generate from landing page copy (e.g., pulling a customer testimonial about "10x revenue" and using it as a headline), consumer-oriented language that does not fit a B2B sale ("easy," "instant," "free" in contexts where your actual sales motion is enterprise), and competitor name mentions that AI might surface from your landing page comparison tables. Set exclusions early, before AI Max generates copy at scale, rather than after you have seen the problem in search term reports.
The configuration sequence: what to set before you enable AI Max
The right order of operations matters. Enabling AI Max before the control layer is in place means the algorithm runs unconstrained during its highest- volume learning phase, which is when poor configurations do the most damage. The recommended sequence for a B2B SaaS account is:
First, set brand keyword exclusions across all campaigns where AI Max will be enabled. Verify that your full brand vocabulary is covered. Second, upload customer match audiences and build your SQL-level remarketing segments — these need time to accumulate match rates before the campaign launch. Third, set URL expansion page inclusions to limit the domain surface AI Max can target — specify only your designated paid landing pages, not the whole domain. Fourth, configure text guidelines with your term exclusions. Then, and only then, enable AI Max. Run the URL-level report on day 14 and day 30 after launch to catch any distribution issues early.
Accounts that follow this sequence are the ones that show up in the minority that reports good results from AI Max. The Dynamic Search Ads to AI Max migration guide covers the parallel track for accounts being automatically upgraded in September 2026 — those accounts should complete this control configuration before the migration date, not after.
What to monitor in the first 90 days
After enabling AI Max with the full control layer in place, the monitoring cadence is weekly for the first 30 days, then bi-weekly from day 31 to 90. The four metrics to track are: search term report quality (are new query expansions matching your ICP intent?), URL distribution (is traffic concentrating on high-converting pages?), brand query leakage (are any brand terms appearing in the AI Max campaign's search term report?), and cost per SQL compared to your standard Search campaigns.
The cost per SQL comparison is the most important. If AI Max is expanding into new query territory, some of those queries will generate form fills that do not convert to SQL — that is expected and not itself a problem. What matters is whether the additional volume produced by AI Max delivers SQL at a comparable or lower cost than standard Search. If yes, expand AI Max to additional campaigns. If cost per SQL is deteriorating despite having the controls in place, the next diagnostic is offline conversion signal quality — the algorithm cannot optimize toward SQL if SQL data is not flowing back to the account on a consistent cadence.