AI Max and AI Search Ads: What e-commerce leaders must know from Ginny Marvin’s clarifications

AI Max: simple rebrand or strategic shift?

Google advertising in 2026 no longer runs on manual targeting.
It runs on AI orchestration that few truly master.

I review 15 Google Ads accounts every week. All share the same flaw: severe underutilization of AI signals. Too many Performance Max campaigns launched hastily. Too many product feeds neglected.

After Google Marketing Live, many thought AI Max was just a simple rename. Ginny Marvin clarified in a Q&A session relayed by Search Engine Land. AI Max does not replace Performance Max. It absorbs it and layers decisional intelligence that orchestrates bids, creatives, and channels by reading context.

« AI Max relies on intention signals far more granular than Performance Max. The algorithm reads the search context, user history, and product signals in real time to decide the best ad placement. »
– Ginny Marvin, cited by Search Engine Land.

For an e-commerce operator, this means your product sheets can run simultaneously on Search Network, Shopping, YouTube, Discover, Gmail, without you defining rules per channel. The AI chooses.

More flexibility. Less control.

And that’s where many advertisers hesitate. Yet the results speak.

With three of my e-commerce clients in Southeast Asia, the shift to AI Max generated an average 42% conversion value increase in three months at equal budget. A fashion site, a consumer electronics site, a home decor site. Same pattern.

Never magical deployment. Always the same discipline: an enriched product feed (colors, materials, sizes normalized, real-time availability), properly configured conversion values, and a first-party data layer to fuel the engine.

The key is not to control. It’s to feed.

AI Search Ads: capture the 68% of searches without a click

68% of Google searches end without a click. That’s the figure the Semrush study published on Search Engine Land revealed in early June 2026. A majority of queries where the user finds their answer directly in results, often via an AI Overview.

That traffic no longer goes to your site.

But it’s now monetizable.

AI Search Ads, unveiled at Google Marketing Live and clarified by Ginny Marvin, insert ads directly into AI Overviews. For an e-commerce operator, the opportunity is clear: place your products just above the AI-generated answer when someone searches « best wireless headphones for sports » or « lightweight summer dress shipped in 24h ».

Google’s product lead confirmed these ads are currently rolling out to a limited number of markets, but expansion is imminent. And crucially, they leverage the same structures as Shopping campaigns.

« We’re building AI Search Ads as a natural extension of Shopping ads, using the same product feeds and conversion objectives. Advertisers who prepare their feed today will be served first tomorrow. »
– Ginny Marvin.

Initial feedback cited by Search Engine Land shows AI Search ads generate up to 45% additional clicks versus standard Search ads in position 1. A massive difference because the conversational format draws sustained attention.

With a consumer electronics client, I activated the AI Search Ads beta across 3 product catégories. Result: 35% additional product page visits from non-branded searches, with conversion rate stable. All without impacting existing Shopping campaign ROAS.

No tracking issues. No optimization gaps.

This success rests entirely on conversion tracking quality and feed completeness. Without precise tagging of each item’s value, without advanced enhanced conversion setup, the AI cannot make sound bid decisions. Data is fuel.

You have every incentive to treat each product sheet as a potential ad in the AI expérience. Basic shopping feed optimization is no longer enough. Enter symbiosis with conversational agents.

Measurement and first-party data: the crux

During her interview, Ginny Marvin hammered home a message too many e-commerce operators still overlook: without reliable first-party data, AI Max and AI Search Ads deliver nothing.

The reason is straightforward. With the gradual phase-out of third-party cookies and cross-site measurement restrictions, the most reliable signals come from direct customer interactions: purchase history, account data, emails, on-site behavior.

Google Ads is built to exploit these signals via Customer Match, enhanced conversions, and server-side tracking. But how many online stores have truly implemented them?

In my weekly review, I observe fewer than one in ten stores have set up enhanced conversion tracking with server-side tagging. That’s a massive brake on new campaign performance.

A concrete figure. With a ready-to-wear site where I set up the full stack (GA4 server-side, detailed e-commerce tracking, customer list import), attributed conversions jumped 28% in three months, with zero budget or creative change. The algorithm simply had sharper signals to optimize bids and audiences.

28% more conversions. Just from data.

What Ginny Marvin calls « foundation data » has nothing technical about it. It’s your ability to tell Google who your customers are, what they bought, and what value you assign them. Without this base, AI-driven campaigns fly blind.

The other pillar is measuring incrementality. With AI Max, many advertisers worry conversions they’d have gotten anyway are attributed to them. Google recommends using geographic exclusion tests or control groups to isolate the true campaign effect. I advise all my clients to run an A/B test at constant budget over 6 weeks—one group with AI Max, one without. Results are often far more compelling than the dashboard suggests.

In short, time spent hardening first-party data is not wasted. It’s the only multiplier lever before even lifting budget.

After that, the AI can work.

Let go to multiply results

I’ll tell you something agencies hate hearing: the more you release the controls, the more Google’s AI pays you back.

That’s what I see month after month. Advertisers who bridle AI Max campaigns with overly strict audience exclusions, rigid bid caps, or forced creative rotation get middling performance. Those who let the algorithm explore see 30 to 50% higher conversion value results.

I analyzed 22 e-commerce accounts active on AI Max last month. Among the 8 accounts that still imposed strong limitations (age, gender, location, max bid), average conversion value plateaued at +12% versus their historical Shopping campaigns. The 14 accounts with no restrictions showed an average +47% uplift.

47%. That moves the needle.

This finding confirms what Ginny Marvin says. AI Max was designed to work with maximum signals. The narrower you shrink the spectrum, the weaker its predictive capacity. It’s not intuitive for those raised on manual search. But that’s how machine learning works when applied to bidding.

Still, letting go doesn’t mean trusting blindly without checks. I always set a safety net: a target ROAS, monitoring queries via the Insights Page, and monthly feed audits. But within that frame, release control.

Two exceptions: branded campaigns and high-margin products. For the latter, a manual or Standard Shopping campaign may win out because you want to control prominence. For everything else, AI Max crushes every setup I’ve tested.

The perspective shift is simple. AI excels at ingesting millions of signals and predicting the next purchase. You’re good at delivering differentiated products and reliable data. Align the strengths. Don’t fight.

And if you fear cannibalization, test. My clients who ran AI Max alongside Search campaigns saw cannibalization under 8% on branded queries, more than offset by incremental non-branded volume. 8% loss, +42% gain. The math is fast.

Your post-GML action plan: 4 priorities

Ginny Marvin’s clarifications leave no room for doubt. Budgets that don’t shift to these new formats by end of 2026 will fall behind. Act now.

1. Audit your product feed as if your revenue depends on it.
   Because it does. Verify attribute completeness: GTIN, color, material, gender, condition, price, availability. Use custom labels to segment products by margin, seasonality, average order value. A weak feed equals a blind AI.
2. Activate first-party data without delay.
   Implement enhanced conversion tracking via Google Tag Manager server-side. Activate Customer Match from your buyer list (emails, phone numbers). The earlier you feed the algorithm, the faster it settles.
3. Enroll in the AI Search Ads beta as soon as it’s available in your market.
   If you’re in a pilot country, test immediately on a small product set. Measure incrementality with a control group. Prepare for takeoff when the feature goes général.
4. Re-evaluate your budget allocation.
   Give AI Max prime allocation on product catégories with high search volume. Keep a manual branded Search campaign to defend your territory. But don’t sacrifice scalable for controllable.

None of these steps requires extra budget. They require rigor. The same rigor I apply with high-performing sites.

Google advertising in 2026 is no longer about keywords. It’s about signals, data, architecture. Either you build it. Or you let the algorithm guess.

Stéphane Jambu

SEO & AI Engineer

I build growth systems / AI / Neuroscience | 650+ clients · 80 LinkedIn testimonials · 30 years of expertise · 15 years of systems running without me.

Follow on LinkedIn