i manage $2.4M in annual ad spend across google and meta for lead-gen businesses. when google announced that dynamic search ads, automatically created assets, and campaign-level broad match would all be force-upgraded to AI Max by september 2026, i had to figure out fast what this means for real accounts with real budgets. Google AI Max is not a new campaign type. it is an AI optimization layer you switch on inside your existing search campaigns. and if you do not configure it correctly, it will waste your budget on irrelevant queries and send traffic to landing pages that do not convert.
this guide covers everything i have learned about AI Max: what it actually does, what changed since launch, the prerequisites your account must meet before testing, what to turn off immediately, how to run experiments without losing control, and the real performance data (not just google’s marketing numbers). whether you run lead-gen for local businesses or manage enterprise search campaigns, this is the AI Max guide i wish existed when i started testing.
the short version: AI Max can work for lead-gen, but only if you disable URL expansion, lock down auto-generated assets on brand campaigns, and run it as an experiment first. do not enable it across your entire account in september just because google is force-upgrading your legacy campaigns.
in this guide:
- what AI Max actually is (and what it is not)
- the three engines: search term matching, text customization, URL expansion
- why september 2026 matters (the force-upgrade deadline)
- prerequisites your account must meet before testing
- what to turn off immediately for lead-gen campaigns
- how to test AI Max with one-click experiments
- AI Max vs Performance Max: overlap and cannibalization
- the real performance data (independent testing vs google claims)
- common failure points and how to avoid them
- FAQ covering brand controls, negatives, and reporting
What Is Google AI Max?
Google AI Max is an AI optimization layer that runs inside your existing search campaigns. it is not a new campaign type. you activate it at the campaign or ad group level, and it upgrades three specific features: search term matching, text customization (formerly automatically created assets), and final URL expansion. once enabled, AI Max processes real-time signals to automatically refine your targeting, ad copy, and landing page selection.
the three engines work together:
1. search term matching
this is the core of AI Max. it uses broad match and keywordless technology to find search queries that your current keyword list misses. instead of matching only the keywords you explicitly add, AI Max learns from your existing keywords, creatives, and landing pages to identify relevant searches across google’s entire query universe. for accounts that have hit a ceiling with their current keyword coverage, this can unlock incremental demand. for accounts with broad match waste, it can amplify the problem.
2. text customization
formerly known as “automatically created assets” (ACA), text customization uses your existing ad copy, landing page content, and generative AI to create new headlines and descriptions tailored to specific search queries. google claims this makes your ads more relevant to each individual search. in practice, the quality varies widely. some generated assets are serviceable. many are generic, off-brand, or make claims your landing page does not support. for lead-gen campaigns where compliance and accuracy matter, auto-generated assets can be a liability.
3. final URL expansion
this feature automatically sends users to the landing page on your domain that google predicts will perform best for each query. instead of sending all traffic to a single designated landing page, URL expansion distributes traffic across your site based on relevance signals. for ecommerce sites with hundreds of product pages, this can work well. for lead-gen businesses with specific conversion pages, it can be catastrophic. if URL expansion sends a user to your blog post instead of your contact form, you lose the conversion.
key takeaway for lead-gen: search term matching is the most valuable AI Max engine for lead-gen. text customization is a mixed bag that requires close monitoring. URL expansion should generally be disabled for lead-gen campaigns unless you have strict exclusion lists in place.
Why September 2026 Matters
google announced three force-upgrade paths that converge in september 2026. if you run any of these legacy configurations, the clock is running:
1. dynamic search ads (DSA)
all DSA campaigns will be upgraded to AI Max with all three features enabled by default: search term matching, text customization, and final URL expansion. your legacy URL controls carry over, but URL expansion will be active unless you explicitly disable it. after september, you will not be able to create new DSA campaigns through the google ads UI, editor, or API.
2. automatically created assets (ACA)
campaigns with ACA enabled will upgrade to AI Max with search term matching and text customization on. final URL expansion is not enabled by default in this path, which is safer for lead-gen. but text customization will be generating new headlines and descriptions without your approval.
3. campaign-level broad match
campaigns with a broad match setting at the campaign level will also upgrade to AI Max. google has not published a complete breakdown of which features activate in this scenario, so the behavior is less predictable.
the september deadline is real. google expects all eligible campaigns to be upgraded by the end of september 2026. if you wait for the automatic upgrade, you inherit a config that mirrors your old setup. the official recommendation is to upgrade proactively so you can tune the controls (brand, text, URL exclusions) from day one.
Prerequisites Your Account Must Meet
not every account is ready for AI Max. independent testing shows that 84% of advertisers who tested AI Max reported neutral or negative results. this is not because AI Max does not work. it is because most accounts are not properly prepared. the 16% who see positive results share common characteristics:
| prerequisite | why it matters |
|---|---|
| accurate conversion tracking | AI Max learns from your conversion signals. if tracking is incomplete or delayed, the AI optimizes for the wrong outcomes. enable enhanced conversions and import offline conversions for lead-gen. |
| comprehensive negative keywords | broad match and keywordless matching expand reach fastest where negatives are weakest. without robust negative lists, AI Max will spend on irrelevant queries. |
| quality landing pages | if URL expansion is enabled, AI Max routes traffic to pages it predicts will perform best. if your site has thin, outdated, or irrelevant pages, URL expansion will send traffic to them. |
| adequate conversion volume | google recommends 100+ monthly conversions for AI Max to have enough signal to learn from. below that, the AI is guessing. |
| proven broad match success | accounts that have already succeeded with broad match keywords are better positioned for AI Max. if broad match burns your budget, AI Max will amplify the problem. |
| conversion-based smart bidding | AI Max requires Target CPA, Target ROAS, Maximize Conversions, or Maximize Conversion Value. manual CPC does not provide the signals AI Max needs. |
before enabling AI Max, audit these three things: conversion tracking accuracy, negative keyword coverage, and landing page quality. fix gaps first, then test AI Max as an experiment.
What to Turn Off Immediately for Lead-Gen
AI Max defaults are aggressive. google enables all three engines when you activate it, and the defaults are optimized for broad reach, not lead quality. for lead-gen campaigns, you should review and likely restrict these settings:
1. disable final URL expansion
for lead-gen businesses, URL expansion is the riskiest AI Max feature. if AI Max sends a user to your blog post, about page, or outdated service page instead of your contact form, you lose the conversion. lead-gen funnels are typically narrow and controlled for a reason. disable URL expansion at the campaign level unless you have configured strict landing page exclusion lists.
if you do enable URL expansion, immediately add exclusion lists for pages you do not want receiving paid traffic: about pages, blog posts, policy pages, out-of-stock pages, support pages, and irrelevant international pages.
2. disable auto-generated assets on brand campaigns
text customization generates headlines and descriptions without your approval. on brand campaigns, auto-generated assets can make claims that do not match your landing page or use messaging that does not fit your brand voice. turn off text customization on brand campaigns where you already convert well.
on non-brand campaigns where you want to test text customization, monitor the generated assets weekly. google provides an asset report that shows how AI Max assets are being optimized. remove any assets that are generic, off-brand, or make unsupported claims.
3. disable auto-apply recommendations
google has been auto-enrolling existing campaigns into AI Max through recommendations and auto-apply settings. check your account-level settings and disable any auto-apply recommendations related to AI Max. you want to control when and how AI Max is enabled, not have google decide for you.
4. add brand exclusions
AI Max can end up serving on your own brand terms, inflating a generic campaign’s apparent efficiency. set brand exclusions on non-brand campaigns to prevent AI Max from capturing branded search that your brand campaign should handle. this prevents cannibalization and gives you cleaner performance data.
5. build search theme exclusions
search theme exclusions allow you to specify topics you do not want AI Max to match against. this is similar to negative keywords but operates at the theme level, giving you a broader blocking mechanism. if you offer “CRM software,” add search theme exclusions for unrelated topics like “free CRM templates” or “CRM jobs.”
How to Test AI Max with One-Click Experiments
never enable AI Max across your entire account simultaneously. google built experimentation directly into the AI Max activation flow, and you should use it. when you enable AI Max on a campaign, you can choose to run an experiment rather than activating it outright. the system creates a 50/50 traffic split between your campaign with AI Max enabled and the same campaign with AI Max disabled. google handles all the statistical analysis, tracking conversion lift, cost efficiency, and confidence intervals.
follow this testing framework:
- pick one campaign: choose your highest-performing non-brand search campaign with at least 100 monthly conversions and established broad match success.
- run the experiment for 4-6 weeks: google recommends a minimum of 4 weeks to allow the learning period to complete and reach statistical significance. do not judge results in the first week.
- measure what matters: do not just look at conversions. track conversion rate, cost per conversion, qualified lead rate (not just leads), and revenue per conversion. AI Max can increase total conversions while decreasing lead quality.
- review search terms weekly: the search terms report is your control tool number one. without a weekly review, you miss the waste. add negative keywords and search theme exclusions based on what actually comes in.
- only scale after clear positive results: if the experiment shows clear positive results after 4-6 weeks, gradually enable AI Max on additional campaigns. if results are flat or negative, fix your prerequisites and retest.
AI Max vs Performance Max: Overlap and Cannibalization
AI Max for search and Performance Max are both google AI products, but they solve different problems and compete for budget in different ways. understanding the distinction is critical for any 2026 google ads strategy.
the key differences
| aspect | AI Max for search | performance max |
|---|---|---|
| what it is | optimization layer inside existing search campaigns | separate cross-network campaign type |
| channels | search only | search, shopping, display, youtube, discover, gmail, maps |
| creative | text customization (headlines, descriptions) | all asset types (text, image, video) |
| control | more granular (search term exclusions, URL exclusions, brand controls) | less granular (audience signals, asset groups) |
| best for | incremental search query expansion | full-funnel, cross-channel coverage |
overlap and cannibalization
AI Max search campaigns and PMax can both claim credit for the same conversions. if a user searches for your brand term and converts, both the AI Max search campaign and the PMax campaign might attribute that conversion. this overlap inflates apparent performance without growing actual revenue. to manage this, set brand exclusions on non-brand AI Max campaigns and ensure your brand campaigns are separate and protected.
when to use which
- use AI Max search when you have strong search campaigns that have hit a ceiling and you want to test incremental query expansion with some guardrails.
- use PMax when you want to reach users across all google inventory with a single campaign and have a large, changing catalog.
- run both when you want broad coverage but need to manage overlap through brand exclusions and budget allocation.
The Real Performance Data
google publishes two different performance numbers for AI Max, and they answer different questions:
| figure | google’s claim | what it actually means |
|---|---|---|
| 14% | “advertisers that activate AI Max typically see 14% more conversions at similar CPA/ROAS” | AI Max activated vs not activated. this is the headline number, but it includes all advertisers, not just those with prepared accounts. |
| 7% | “campaigns see 7% more conversions when using the full feature suite vs search term matching alone” | Full AI Max (matching + text + URL) vs search term matching only. this is the measured price of losing control over ad copy and landing pages. |
independent testing tells a more nuanced story:
- 84% of testers report neutral or negative results (Source: independent analysis across hundreds of accounts, early 2026)
- 99% of impressions across ~30,000 AI Max search terms produced zero conversions (Source: August 2025 analysis)
- 35% lower ROAS vs traditional match types across 250+ retail campaigns (Source: November 2025 study)
these numbers do not mean AI Max cannot perform. they mean that the quality of your inputs determines the quality of your results. accounts with clean conversion tracking, robust negatives, and strong landing pages see the best results. accounts without these foundations see waste.
realistic expectations for well-prepared accounts: 10-20% conversion improvement at similar CPA. achieving 10% puts you ahead of the vast majority of testers. for accounts with some readiness gaps, expect flat performance with incremental reach. for unprepared accounts, expect higher CPAs and wasted spend.
Common Failure Points and How to Avoid Them
most AI Max failures come from a handful of preventable mistakes. here are the ones i see most often:
1. enabling AI Max on thin or outdated sites
if your website has many poor pages (thin content, outdated service pages, old blog posts), URL expansion will send traffic to them. before enabling AI Max, audit your site and either improve or exclude pages that should not receive paid traffic.
2. weak or immature negative keyword lists
broad match and keywordless matching expand reach fastest where negatives are weakest. if you do not have comprehensive negative keyword lists, AI Max will spend on irrelevant queries. build your negatives first, then test AI Max.
3. ignoring the search terms report
the search terms report is your control tool number one. without a weekly review, you miss the waste. add negative keywords and search theme exclusions based on what actually comes in, not what you expect to come in.
4. judging results too early
every material change resets the learning period. give AI Max 4-6 weeks to learn before judging results. do not turn it on and off repeatedly, as each reset restarts the learning clock.
5. not separating brand and non-brand campaigns
AI Max can capture branded search on non-brand campaigns, inflating apparent performance. separate brand and non-brand campaigns, and set brand exclusions on non-brand campaigns to prevent cannibalization.
6. waiting for the automatic upgrade
if you wait for the september force-upgrade, you inherit a config that mirrors your old setup. upgrade proactively so you can tune the controls (brand, text, URL exclusions) from day one.
Frequently Asked Questions
is AI Max a new campaign type?
no. AI Max is an optimization layer that runs inside your existing search campaigns. it is not a separate campaign type. you activate it at the campaign or ad group level, and it upgrades search term matching, text customization, and final URL expansion. Performance Max is a separate cross-network campaign type that runs across search, shopping, display, youtube, gmail, and maps.
should I enable AI Max for lead generation?
yes, but only if you disable URL expansion and lock down auto-generated assets on brand campaigns. search term matching is the most valuable AI Max engine for lead-gen. text customization is a mixed bag that requires close monitoring. URL expansion should generally be disabled for lead-gen campaigns unless you have strict landing page exclusion lists. run AI Max as an experiment first, not across your entire account.
what is the deadline for the AI Max upgrade?
september 2026. google expects all eligible campaigns (DSA, ACA, campaign-level broad match) to be upgraded by the end of september 2026. after september, you will not be able to create new DSA campaigns. google recommends upgrading proactively so you can configure the controls from day one.
does AI Max work with manual CPC?
no. AI Max requires conversion-based smart bidding. you need Target CPA, Target ROAS, Maximize Conversions, or Maximize Conversion Value. manual CPC does not provide the signals AI Max needs. switch your bid strategy first, then test AI Max.
how do I control what queries AI Max matches?
through negative keywords and search theme exclusions. AI Max treats all keywords as broad match regardless of how you originally configured them. there is no toggle to prevent the broad match behavior directly. your defense is robust negative keyword lists, search theme exclusions, and weekly search term report reviews.
can AI Max send traffic to the wrong landing page?
yes, if final URL expansion is enabled. URL expansion automatically sends users to the page google predicts will perform best for each query. for lead-gen, this can send users to blog posts or about pages instead of your contact form. disable URL expansion at the campaign level, or configure strict landing page exclusion lists.
what is the difference between AI Max and AI Brief?
AI Max is the optimization layer. AI Brief is a natural language input layer inside AI Max. powered by Gemini, AI Brief lets you write a short brief in plain language to steer AI Max. you set messaging guidelines (what ads should and should not say), matching guidelines (which queries to target or avoid), and audience guidelines (who you are trying to reach). AI Brief is rolling out in english for search campaigns first.
should I test AI Max or wait for the automatic upgrade?
test now, on your own terms. if you wait for the september force-upgrade, you inherit a config that mirrors your old setup. upgrading proactively lets you tune the controls (brand, text, URL exclusions) from day one. use the built-in one-click experiment feature to test on a single campaign before committing.