AI Max for Search: +14% conversions with no extra work?

And what does the independent data say? A Smarter Ecommerce study that analysed more than 250 retail campaigns found that while the median revenue lift was +13%, CPA rose by 16% at the same time. And ROAS? Unchanged. The spread of results is enormous — from +42% to -35%. The study's authors summed it up dryly: “Essentially a coin toss.”

So — switch it on, or not? It depends on exactly what you do and what you expect from AI Max. In this article we'll show you what AI Max actually does, where it works brilliantly, where it fails, and how to test it without putting your budget at risk.

What AI Max for Search (and what it isn't)

First, let's be clear about what AI Max actually is, because there's a lot of confusion around it. AI Max for Search is not a new campaign type like Performance Max. It's a set of features you switch on inside an existing Search campaign. With a single toggle.

What happens once you turn it on? Three things:

1. Search-term expansion (Search Term Matching)

The AI looks at your keywords, ads and landing pages and starts showing ads for queries you never added manually. It works much like broad match, but goes further — the AI can serve an ad with no link to your keywords at all if it predicts a conversion.

2. Dynamic ad text (Text Customization)

The AI generates headlines and descriptions from your website content, existing ads and keywords. It can even rewrite pinned assets if it predicts better performance. That's important to grasp — AI Max can change the copy you crafted so carefully.

3. Final URL expansion (Final URL Expansion)

The AI routes users to whichever pages on your site it considers most relevant — even if that isn't the page you set in the ad. Similar to the old Dynamic Search Ads. And note: it's on by default. If you don't want it, you have to turn it off manually.

How does AI Max differ from Performance Max?

This is the question we get asked most often. The short summary:

In short: AI Max adds an AI layer to your Search campaigns but leaves you in control. PMax does everything itself across every channel. They aren't competitors — they complement each other.

What the data says: where AI Max works and where it doesn't

Here we need to be honest. Google's numbers look great, but the reality is more complicated.

Where AI Max works brilliantly

Lead generation and services with limited search volume. Here the data is unambiguous:

• ClickUp (productivity SaaS): +20% incremental conversions, +16% ROAS, -22% CPA
• Royal Canin (premium pet food): +263% conversions, -73% CPA
• Klook (travel platform): +161% conversion value, +25% ROAS in one month
• L'Oréal: 2× higher conversion rate, -31% cost per conversion
• MyConnect: +16% leads, -13% cost per lead, with 30% coming from new queries

Why does it work? Because these companies had limited search volume, and AI Max opened up new queries they would never have targeted by hand. For niche services with a few dozen conversions a month, AI Max can deliver exactly the extra volume that's missing.

Where it's a “coin toss”

E-commerce and retail. Here the data is far less rosy:

The SMEC study (250+ retail campaigns) showed:

• Median revenue lift: +13%
• Median CPA increase: +16% (new conversions cost more)
• Median change in ROAS: 0%
• Spread: from +42% to -35%
• Only 22% of campaigns hit their original ROAS target

Another study analysed 23 tests across 16 advertisers. The result? At the campaign level, +7% conversion value from new queries — but at the account level, just +3% of net incremental uplift. Why? Because 54% of the “new” queries were already covered by other campaigns in the account.

One agency reported costs “two to three times higher than traditional Search campaigns”, and the +14% claim did not hold up in their tests.

Why such a difference?

Lead-gen firms typically have a limited pool of search queries, and AI Max expands it. Online stores, by contrast, usually already cover a wide range of queries through Shopping, PMax and broad match Search. AI Max then adds queries that are only marginally relevant — and you pay more for them.

Text Guidelines: how to steer the AI so it doesn't damage your brand

In our view, this is the most important development of 2026 for anyone using or considering AI Max. Text Guidelines have been globally available since February 2026 and change the way you control AI-generated copy.

You have two tools:

Term Exclusions — up to 25 specific words or phrases the AI may not use in ads. A competitor's name, sensitive terms or misleading claims, for example.

Messaging Restrictions — up to 40 rules written in natural language. This is where it gets interesting. You can write:

• “Never mention discounts above 30%”
• “Always mention free shipping over €40”
• “Write formally, don't use slang”
• “Don't use superlatives like ‘best' or ‘cheapest'”
• “Always mention the 30-day money-back guarantee”

The rules work across all of a campaign's languages — even if you write them in one language, the AI applies them to ads in every other language too.

A real-world example: carmaker BYD set Text Guidelines for its AI Max campaigns and achieved +24% leads at -26% cost. The key was precisely defining messaging restrictions and brand-voice rules.

Without Text Guidelines, AI Max is risky — the AI can write anything it thinks will convert. With Text Guidelines, you're in control. That's why we recommend: set your rules before you switch AI Max on.

How to test AI Max without the risk

If AI Max interests you, here's how to trial it safely:

Step 1: Pick a campaign with enough data

Ideally a campaign with 50+ conversions a month, stable performance and a Smart Bidding strategy. Don't test on new campaigns with no history.

Step 2: Set Text Guidelines BEFORE switching on

Define your messaging restrictions and term exclusions. What the AI must not say, what tone it should use, what information it must always include. This is your safety net.

Step 3: Run an experiment (50/50 split)

In Google Ads, create an experiment where half your traffic runs with AI Max on and half without. Don't switch AI Max on across whole campaigns blind — you'll have nothing to benchmark against.

Step 4: Exclude branded queries

AI Max tends to serve ads on branded searches where you'd win the conversions anyway. Exclude brand terms so you're measuring the real incremental gain.

Step 5: Review the search terms report every week

This is critical. AI Max generates a huge volume of new queries — sometimes tens of thousands of rows. Go through them and add negative keywords. Without that, you'll be paying for irrelevant traffic.

Step 6: Evaluate after 3–4 weeks

Compare the experiment group against the control. Look at conversions, CPA and ROAS — but at the account level, not just one campaign. AI Max can “steal” conversions from other campaigns.

TIP: If your account already runs a lot of broad match keywords or active Dynamic Search Ads, AI Max probably won't deliver a meaningful improvement — you already cover those queries. The biggest gains come for accounts built mainly on exact and phrase match.

AI Max in the context of the Power Pack strategy

At Marketing Live 2025, Google introduced the so-called Power Pack — a trio of campaign types designed to cover the entire purchase funnel:

• Demand Gen → building awareness and interest (top of the funnel)
• AI Max for Search → capturing active interest when people are searching (the middle)
• Performance Max → conversions across Google's entire network (the bottom)

It's the vision Google is steering the whole platform toward. In practice, it means AI Max isn't a replacement for PMax — it's a complement. PMax covers Shopping, Display, YouTube and other channels. AI Max expands Search.

For online stores, running both makes sense — PMax for the product feed and visual formats, AI Max to expand text-based search campaigns. But only if you have the budget and conversion volume for both to get enough data.

At LK Media, we test the Power Pack approach with clients spending more than €4,000 a month. For smaller budgets, we recommend focusing on one or two campaigns rather than fragmenting your data too thin.

Conclusion

AI Max for Search is an interesting tool, but it's no miracle. For lead generation and niche services with limited search volume, it can deliver a marked improvement. For online stores, the results are inconsistent and demand careful testing.

The key: don't just listen to Google. Test on your own data, set Text Guidelines, monitor your search terms and evaluate at the account level. And if the numbers don't add up, simply switch AI Max off — it's a toggle, not a lifelong commitment.

Want help testing AI Max in your account? Email us at info@lkmedia.cz — we'll gladly set up the experiment and assess whether it makes sense for your business.

FAQ