Google AI Max for Shopping Campaigns: Feed-Driven Setup Guide for Ecommerce

What AI Max for Shopping Actually Changes

AI Max started as a Search campaign feature — expanding keyword matching and enabling final URL replacement. In Q1 2026, Google extended it to Shopping campaigns, and the underlying logic shifted in a meaningful way.

Instead of matching your manually entered keywords, Gemini now reads your Merchant Center product feed and matches products against conversational user queries. A user searching “lightweight sleeping bag for camping in sub-zero temperatures under $300, a few recommendations” — a query no keyword list covers cleanly — gets served your product if your feed describes the temperature rating, weight, and price accurately.

For ecommerce advertisers, the leverage point moved. Keyword coverage mattered when you were manually buying terms. Feed quality matters now. What Gemini can’t read in your feed, it can’t match.

Enabling AI Max for Shopping

Open an existing Standard Shopping campaign or create a new one in Google Ads. Look for the AI Max toggle in campaign settings. As of Q2 2026, the feature is generally available to Shopping advertisers, though some accounts may still need to enable it under Feature Preview in account settings.

Three prerequisites to confirm before flipping the switch:

Merchant Center version. AI Max for Shopping requires Merchant Center Next. Most accounts migrated in late 2025, but manually managed legacy accounts may still be on the old version. Check under Settings in Merchant Center.

Bidding strategy. AI Max doesn’t support manual CPC. Switch to a smart bidding strategy — Target ROAS or Target CPA — before enabling. If you’re currently on manual bidding, make that switch first and let the algorithm accumulate a few weeks of data before adding AI Max.

Campaign scope. Start with one sub-category or a seasonal product line rather than your full catalog. Run two weeks of data before deciding whether to expand.

AI Brief: Steering Gemini With Plain Language

AI Brief is a new input layer inside AI Max settings that addresses the “black box” complaint most advertisers had about early AI Max behavior. Before AI Brief, you had limited visibility into why certain products were triggering certain queries. AI Brief gives some of that control back.

Access it through the AI Max section of campaign settings. It presents three structured text fields:

Brand context. Describe what you sell, who buys it, and your core differentiator. Write it conversationally — you’re briefing an AI, not writing ad copy. For example: we sell outdoor camping gear, primary customers are weekend campers aged 35-55 in the US, not professional mountaineers, price range $50-300.

Message focus. Tell Gemini what you want the ads to communicate and what to avoid. For example: lead with lightweight and fast pack-down, never mention competitor names, don’t imply discount positioning — we don’t discount.

Audience guidance. Describe audiences you don’t want. For example: skip B2B purchasing traffic, not a good fit for students given the price point, focus on repeat buyers and high-value new customers.

AI Brief inputs act as system-level guidance. They don’t guarantee exact behavior, but tested results show noticeably more predictable targeting compared to campaigns without a brief — particularly in categories where brand tone is sensitive, such as pet nutrition or baby products.

Feed Fields That Determine AI Max Performance

AI Max for Shopping uses feed data to match queries. These are the fields with the highest impact:

Feed fieldAI Max dependencyOptimization direction
titleHighPut the most important attributes in the first 30 characters — brand, product type, key spec. No keyword stuffing.
descriptionHighDescribe specific use cases and functions. Avoid generic language.
product_typeMediumUse your own category taxonomy, more granular than Google’s default
custom_label 0-4MediumTag by season, price tier, margin — supports layered bidding
color / size / materialHighColor and material are high-frequency filters in conversational queries
GTINHighProducts without GTINs get lower AI matching priority

Title is the highest-leverage field. Most feed titles are pulled directly from product pages, formatted for SEO: “Black Waterproof Jacket XL.” AI Max needs titles that match natural language queries. “Waterproof packable rain jacket, lightweight hiking camping, men’s” performs substantially better for conversational matching.

Common Setup Mistakes

Short-term impression drop. AI Max goes through a learning period of 2-3 weeks while it processes your feed and accumulates conversion data. Don’t compare daily impression volume against the pre-AI Max baseline during that window. The comparison is meaningless.

AI Max and Standard Shopping competing with each other. Both campaign types bid on overlapping inventory. Run them as separate campaigns, use exclusions to keep branded and exact product queries in Standard Shopping, and let AI Max handle long-tail expansion. Overlap without segmentation wastes budget.

Missing negative keywords. AI Max casts a wide matching net. Add negative keywords at the campaign level before enabling — categories you don’t want, competitor-association queries, adjacent verticals. Fixing this after the fact means spending through unwanted impressions first.

Once data stabilizes, compare the two campaign types on new customer acquisition cost and ROAS. AI Max typically wins on long-tail new customer cost. Standard Shopping wins on branded and exact-match efficiency. Both earning their budget is the right outcome.

Related Articles