Google Ads Rolls Out 1-Click 'AI Max' 50/50 Campaign Experiments With Ad-Group Controls

Google Ads has updated its platform with a new 1-Click “AI Max” Experiment button on the Recommendations page—allowing business owners and digital marketers to instantly launch a 50/50 split test pitting their existing Google Search campaigns against Google’s AI Max automation—while simultaneously adding much-needed ad-group level switches so advertisers can turn off AI keyword expansion or AI-rewritten ad copy where they don’t want it.

For anyone spending money on Google Search Ads to bring in leads or online sales, Google’s “AI Max” search features have been a source of both curiosity and anxiety. On one hand, AI Max promises to catch new conversational searches people type into Google by expanding beyond your exact keyword list and dynamically rewriting your ad headlines and landing page links. On the other hand, veteran marketers know that flipping on full AI automation across an entire campaign all at once can cause Google to spend your budget on irrelevant searches or write generic ad copy that hurts conversion rates.

Previously, setting up a proper scientific test to see if AI Max actually beats your normal Search campaign required building a custom campaign experiment from scratch. With the newly documented update, eligible Google Ads accounts now see a Create 1-click experiment automatically card right inside their Recommendations tab, which immediately spins up a clean 50/50 split test comparing your untouched base Search campaign (the control arm) against an AI Max version (the treatment arm).

Even more useful for hands-on marketers is a second update inside the Experiments tab: instead of being forced to test all of AI Max across every ad group at once, you can now toggle Search term matching and Asset optimization on or off for individual ad groups during experiment setup.

Fast Facts: What Changed in Google Ads AI Max Experiments

• Update & Source: Added to Google Ads’ official AI Max Search FAQ documentation, spotted by PPC analyst Hana Kobzova and detailed in Search Engine Roundtable’s Google Ads analysis.
• How the 1-Click Recommendation Works: Eligible accounts now see a 1-click experiment recommendation on their Google Ads Recommendations page; clicking Create 1-click experiment automatically launches a 50/50 split test comparing the base Search campaign against an AI Max treatment arm with zero manual setup.
• New Ad-Group Level Controls: When building an AI Max experiment manually under the Experiments tab, advertisers can now turn Search term matching (broadening queries beyond exact/phrase keywords) and Asset optimization (AI-rewritten headlines/URLs) on or off for specific individual ad groups.
• Control vs. Treatment Integrity: While the treatment arm tests your chosen AI Max settings on half the traffic, the control arm preserves your exact original Search campaign settings on the other 50% so you can compare real Cost Per Acquisition (CPA) and Return on Ad Spend (ROAS).

Option Comparison: 1-Click Recommendation vs. Custom Experiments Tab Setup

Before you click the shiny new 1-click button on your Google Ads Recommendations screen, it helps to know the difference between the instant 1-click setup and taking two extra minutes in the Experiments tab.

Why the New Ad-Group Toggles Matter So Much for Real Businesses

To see why the new ad-group level controls under the Experiments tab are such a big deal, imagine a local plumbing company, a law firm, or a B2B software brand running a single Google Search campaign with three different ad groups:

1. A “Brand Name” Ad Group (bidding on your own company name);
2. A “High-Intent Exact Price/Service” Ad Group (bidding on expensive $50-per-click keywords like “emergency commercial roof repair”);
3. A “General Research” Ad Group (bidding on broader problem searches).

If you click the 1-click recommendation button blindly, Google enables both Search term matching (letting AI match your ads to looser, related searches) and Asset optimization (letting AI rewrite your headlines and pick different landing page URLs on your website) across all three ad groups for 50% of your budget.

In real life, that creates two common problems:

• First, you almost never want AI “Search term matching” turned on inside a Brand Name ad group or a tightly regulated ad group, because it will bleed expensive clicks into unrelated competitor or informational searches.
• Second, if both Search term matching and Asset optimization are turned on at the same time and your sales go up (or tank), you have no idea which feature caused it. Was it the new search queries AI Max found, or was it the new headlines AI Max wrote?

By going into the Experiments tab instead and using the new granular ad-group controls, you can leave your Brand ad group completely untouched, turn on only Search term matching in your General Research ad group, and turn on only Asset optimization in a separate test—getting clean data you can actually trust.

Practical Checklist Before Running a Google Ads AI Max Split Test

If you manage Google Ads for your own business or for clients, follow these four steps to test AI Max safely this month.

The 4-Step Enterprise Implementation Playbook

1. Use the Experiments Tab Instead of the 1-Click Button If You Mix Brand & Non-Brand Ad Groups: When you see the new 1-click AI Max card on your Recommendations page, treat it as a signal that your campaign is eligible—then click over to the Experiments tab instead so you can exclude brand, competitor, and compliance-locked ad groups before launching the 50/50 split.
2. Test Search Term Matching and Asset Optimization Separately: Do not flip both switches on at the same time in the same ad group if you want to know what works. Enable Search term matching first while keeping Asset optimization off so you can see whether Google’s AI query expansion brings in profitable new search terms without changing your proven landing pages and ad copy.
3. Lock Your Final URL Exclusions Before Testing Asset Optimization: Because AI Max’s Asset optimization can dynamically send clicks to different pages on your website (similar to Dynamic Search Ads / Performance Max), always add URL exclusions for your /blog, /careers, /terms, /login, and /about pages before starting the experiment so you never pay $10 a click to send a shopper to your privacy policy.
4. Let the 50/50 Split Run for at Least 2 to 4 Conversion Cycles and Check the Search Terms Report Weekly: Don’t declare a winner after 48 hours. Run the 50/50 split for 14 to 28 days, and check the Search Terms report inside the AI Max treatment arm every Friday to add negative keywords for any low-intent queries the AI matched.