Google AI Max : An Honest Review After Independent Testing

Google AI Max: An Honest Review After Independent Testing

Google says AI Max for Search campaigns delivers 14% more conversions at similar cost per acquisition. That number gets repeated at every conference, in every product update, and in every case study Google publishes. The problem is that real-world results across hundreds of independent tests tell a more complicated story.

This is a candid breakdown of what AI Max actually does, what the independent data shows, and the specific conditions under which it works — based on testing across multiple agencies and a meta-analysis of 250+ retail campaigns.

What AI Max Actually Is

AI Max isn't a new campaign type. It's an optimization layer that sits on top of existing Search campaigns and combines three features:

• Search term matching — broader query coverage using keywordless targeting alongside your existing keywords

• Text customization — automatic generation and modification of ad copy based on user intent signals

• Final URL expansion — dynamic landing page selection where Google's AI picks which page to send traffic to

Google announced AI Max for Search in May 2025 and rolled it out globally in September. In April 2026, Google confirmed it's also retiring Dynamic Search Ads, with auto-migration to AI Max beginning in September 2026.

The pitch is straightforward: use Gemini's understanding of intent to capture searches your existing keyword list misses. The reality varies dramatically by account.

What Google's Numbers Say

Google's headline statistics, frequently cited:

• +14% conversions at similar CPA on average

• +27% conversions for accounts that previously ran mostly exact and phrase match

• +8% higher ROAS from Performance Max compared to Search-only strategies (Nielsen analysis)

• +15% higher ROAS from AI-powered Broad Match compared to other match types

Published case studies include L'Oréal achieving 2x higher conversion rates with 31% lower cost-per-conversion, ClickUp seeing 161% higher conversion value within one month, and Royal Canin capturing long-tail queries previously unreachable through standard keyword targeting.

These are real results from real brands. But they're also case studies — directional signals from advertisers with specific starting conditions, not universal promises.

What Independent Testing Shows

The Smarter Ecommerce study by Mike Ryan analyzed 250+ retail campaigns running AI Max. The findings:

• +13% median revenue uplift — close to Google's claim

• +16% median CPA increase — Google's marketing materials don't highlight this

• ROAS range from +42% to -35% — meaning some accounts saw dramatic improvement, others saw substantial decline

That ROAS range is the critical data point. Same feature, same platform, same time period — wildly different outcomes per account.

Other independent findings:

• A 4-month head-to-head test by Xavier Mantica found AI Max cost 128% more per conversion than phrase match and 91% more than exact match.

• Up to 63% of AI Max's "expanded" queries were searches the account's existing keywords already covered. Not new customers — same customers at higher prices.

• One account experienced AI Max scaling so aggressively into competitor brand terms that it consumed 69% of total Search impressions.

• One campaign saw half a million monthly impressions land on the Search Partner Network at a 0.07% conversion rate, compared to 3.04% on standard Google Search.

• About 16% of advertisers globally have tested AI Max as of early 2026, and few have committed full budgets.

A separate 30-day case study by Coalition Technologies showed a 70.15% revenue increase and 72.01% ROAS boost for an ecommerce client — but the same agency reported poor performance on lead-gen accounts with the same feature.

Where AI Max Actually Works

Across the data, AI Max performs best under specific conditions:

Strong existing conversion data. AI Max relies on the account's historical conversion signals to train. Sites with sparse conversion data don't have enough signal for the AI to optimize meaningfully.

Accounts previously running narrow match types. The biggest improvements come from accounts heavy on exact and phrase match keywords, where AI Max can genuinely expand reach. Accounts already running broad match see smaller incremental gains.

Ecommerce with diverse product catalogs. Multiple landing pages and product types give AI Max more to work with. Single-product or service businesses see less benefit.

Long-tail query opportunities. When users search in conversational, descriptive language ("how many times a day to feed a puppy" rather than "puppy food"), AI Max captures intent that exact match misses.

Strong negative keyword foundation. AI Max amplifies whatever signal you give it. Without disciplined negative keyword management, broader matching becomes broader waste.

Where AI Max Fails

Equally well-documented failure conditions:

Lead-gen accounts. Multiple agencies report AI Max performs significantly worse for lead generation than for ecommerce. The conversion signal is weaker, the customer journey is longer, and AI Max often surfaces queries that look relevant but produce unqualified leads.

Compliance-heavy industries. AI Max's text customization generates ad copy that may not match required regulatory language for healthcare, finance, legal, or pharmaceutical advertisers. Auto-generated copy can create compliance issues that take more time to fix than they save.

Accounts with sophisticated existing setups. When an account already runs modern campaign structures with broad match, automated bidding, and well-developed long-tail keyword coverage, AI Max has less room to add value. Independent tests consistently show smaller (or negative) lifts in these conditions.

Brand campaigns. AI Max can scale into competitor brand terms aggressively, which is often desirable in some accounts but actively damaging in others. Without careful exclusions, you can end up cannibalizing your own branded search traffic.

Accounts without strong conversion tracking. AI Max amplifies the signal you give it. If your primary conversion event is wrong, AI Max optimizes for the wrong outcome faster and at greater scale.

The Cannibalization Problem

The most-reported issue across independent tests is query cannibalization. AI Max's keywordless expansion frequently surfaces queries that existing keywords were already capturing.

In one analysis, the overlap rate reached 63% — meaning more than half of "new" queries were already covered. The advertiser ends up paying additional fees on traffic that was already converting through standard match types, while incremental new query discovery is much smaller than Google's case studies suggest.

This is why measuring AI Max requires careful experimentation, not just toggling it on and watching aggregate metrics.

How to Test AI Max Without Burning Budget

If you want to test AI Max for your account, here's the responsible approach:

• Run an AI Max experiment using Google Ads' built-in A/B test framework rather than enabling it directly on your main campaign. This gives you a clean comparison.

• Allocate 5-10% of search budget to the test for at least 30 days. Smaller percentages risk insufficient data for statistical significance.

• Set text guidelines before launching. Without them, Google generates ad copy with zero brand input.

• Exclude branded search terms to prevent cannibalization of traffic you'd otherwise capture cheaper.

• Monitor search term reports weekly. Track what percentage of "new" queries actually overlap with your existing keywords.

• Add aggressive negative keywords as you identify irrelevant queries. AI Max amplifies whatever you don't exclude.

• Compare CPA, conversion quality, and post-conversion behavior — not just conversion volume. Volume can rise while quality drops.

Document everything before and after. The accounts that get value from AI Max are the ones that treat it as an experiment with clear hypotheses, not a button to push and forget.

Should You Use AI Max?

The honest answer is: it depends on your account.

Use AI Max if: You run ecommerce, have strong conversion data, mostly use exact and phrase match keywords, have well-developed negative keyword lists, and can dedicate someone to monitoring it weekly.

Skip AI Max if: You run lead generation campaigns, operate in a compliance-heavy industry, already use sophisticated broad match with strong existing performance, lack the resources to monitor expanded queries weekly, or rely heavily on brand campaigns you don't want cannibalized.

Plan for AI Max regardless if: You currently run Dynamic Search Ads, automatically created assets, or campaign-level broad match. Auto-migration to AI Max begins September 2026 — manual migration before then gives you more control over how settings transfer.

The Bigger Picture

AI Max is part of a larger pattern across every major ad platform. Manual controls — keyword targeting, audience selection, bid strategies — are being replaced by AI-driven systems month by month. Whether the AI is making you money or making the platform money depends on the inputs you set and the verification you run.

Google's 14% headline isn't wrong. It's just incomplete. The full picture includes the accounts where AI Max produces dramatic gains and the accounts where it produces equally dramatic losses. The advertisers who succeed are the ones who understand both possibilities and test methodically rather than trusting the press release.

FAQ

Does AI Max replace Dynamic Search Ads? Yes. Google announced in April 2026 that DSA, automatically created assets, and campaign-level broad match are being deprecated. Auto-migration to AI Max begins September 2026. Manual migration is recommended now to preserve your settings.

What's the difference between AI Max and Performance Max? Performance Max is a fully automated cross-channel campaign type covering Search, Display, YouTube, Discover, Gmail, and Maps. AI Max is an optimization layer specifically for Search campaigns, with more transparency and control. AI Max stays inside the Search network; PMax doesn't.

How long should I test AI Max before deciding? Minimum 30 days, ideally 60-90 days. Allocate 5-10% of search budget to a structured experiment using Google Ads' A/B test framework. Look at CPA, conversion quality, and post-conversion behavior — not just volume.

Why does AI Max increase CPA in some accounts? Because the broader query matching frequently surfaces searches your existing keywords already covered (up to 63% overlap in tested accounts). You pay for additional impressions on traffic you were already capturing cheaper through targeted match types.

Should I turn on text customization? Only if you've set text guidelines first. Without guidelines, Google generates ad copy with no brand input — which can produce off-brand or factually incorrect ads. Text guidelines act as guardrails for what AI Max can and can't say in your ads.