Google Merchant Center now shows how products are being discovered in AI-powered shopping results, but the report is only useful when the product data behind it is accurate. Google announced in September 2026 that AI performance insights are generally available for eligible Merchant Center accounts in the United States, Canada, Australia, India, and New Zealand. The report covers conversational shopping queries in AI Mode and AI Overviews, including share of voice, shopping stages, popular attributes, and the terms shoppers are prioritizing.

For a small retailer, this is not a reason to chase every AI trend. It is a reason to make the product catalog easier for Google and customers to understand before holiday traffic gets expensive. Clean titles, accurate prices, useful images, reliable availability, shipping details, and consistent landing pages still do the heavy lifting.

What are Merchant Center AI performance insights?

The AI performance report is designed to show how a brand and its products appear for conversational shopping queries across Google AI Mode and AI Overviews. Instead of looking only at short keywords, a merchant can see patterns in the longer, more specific language shoppers use when comparing products and deciding what to buy.

Google’s documentation says the report can include:

• Your share of voice: the portion of eligible AI impressions captured by your brand or products compared with a defined competitor set.
• Competitor average share: a benchmark for whether your visibility is above or below the comparison group.
• Shopping stages: signals that help distinguish discovery, consideration, and decision-oriented behavior.
• Top terms and attributes: the qualities shoppers emphasize in conversational searches, such as material, fit, use case, comfort, compatibility, or durability.

Eligible merchants can look under Analytics > Products > AI performance. Availability is currently limited to English-language queries in supported countries, so the absence of the tab does not automatically mean an account is broken.

Why this matters before holiday shopping accelerates

Holiday shopping queries are rarely as neat as a spreadsheet column. A customer may ask for a gift for a person with a specific hobby, a product that fits an unusual space, or an item that meets several constraints at once. Those searches reward catalogs that explain products clearly rather than relying on a thin title and one generic image.

Google says merchants that adopt core Merchant Center feed best practices see an average conversion increase in the following month. That is Google’s aggregate result, not a guarantee for an individual store. The practical point is stronger: product data cleanup affects free listings, Shopping ads, AI-powered discovery, and the quality of the landing-page experience at the same time.

Start with the Merchant Center “Needs attention” tab

Before interpreting an AI visibility graph, open Products > Needs attention. This section, previously called Diagnostics, shows warnings, disapprovals, and setup or policy issues that can limit or stop products from appearing.

Separate warnings from disapprovals

A warning may allow a product to continue showing with limited performance. A disapproval stops the affected product from appearing until the issue is corrected and the product is re-reviewed. Account-level problems are more serious because they can affect the entire catalog.

Download the affected-item report instead of fixing products from memory. Group issues by cause, count the products involved, and start with errors that remove products from visibility or create a direct mismatch with the website.

Fix the source, not just the symptom

If WooCommerce, Shopify, a spreadsheet, or a feed plugin supplies the catalog, correct the original source whenever possible. A manual edit in Merchant Center may be overwritten during the next sync. Document which system owns the title, price, image, GTIN, shipping value, and availability field so the same error does not return tomorrow.

A practical product feed cleanup checklist

1. Match price and availability exactly

The price and stock status in the feed should match what Google finds on the product landing page. A mismatch can trigger a preemptive item disapproval because Google does not want to send a shopper to an offer that changed after the listing was created.

Check sale prices, scheduled promotions, variable products, backorders, and currency settings. Also confirm that caching does not leave an old price on the page after the store has already sent a new value to Google.

2. Rewrite titles for real identification

A useful product title identifies the item without turning into a pile of keywords. Include the most decision-relevant attributes that distinguish the product, such as brand, model, product type, size, color, or material, when those details are accurate and important.

Do not add promotional language like “best,” “free shipping,” or “limited-time deal” to the title field. Keep those messages in the appropriate promotion, shipping, or sale-price settings.

3. Use complete, factual descriptions

The description should explain what the product is, what it does, and the features a buyer would reasonably need to compare it. Remove leftover HTML, JavaScript, template fragments, and unrelated boilerplate. Product data should describe the product, not the theme or the storewide marketing campaign.

4. Verify product identifiers

Submit a valid GTIN when the manufacturer assigned one. Use brand and MPN where appropriate. Do not invent a barcode to make an error disappear. Truly custom or handmade products can be identified correctly without a GTIN, but the feed needs to represent that situation honestly.

5. Improve primary product images

The main image should show the actual product clearly. Google documents minimum dimensions of 100 by 100 pixels for most products and 250 by 250 pixels for apparel, but minimum compliance is not the same as strong merchandising. Use a large, sharp image where the product occupies most of the frame.

Avoid watermarks, promotional text, fake borders, placeholders, and a logo used in place of the product. Add useful alternate views as additional images so shoppers can inspect scale, details, packaging, texture, or installation.

6. Confirm shipping and return settings

Shipping speed, cost, and return terms are part of the offer. Review the shipping policy table, warehouse locations, cut-off times, transit estimates, and country coverage. Then connect an accurate return-policy page and confirm the Merchant Center settings agree with the website.

7. Test the landing page on mobile

Every listing should lead to the specific product, not a category page or a generic search result. Test the URL without being logged into the store. Check for server errors, blocked images, intrusive popups, broken variants, hidden add-to-cart controls, and slow-loading scripts.

How to use AI shopping terms without ruining the catalog

The new report can reveal attributes that matter in conversational searches. That does not mean every popular phrase belongs in every title.

Map an attribute to evidence

If shoppers prioritize “solid wood,” confirm the product is actually solid wood and identify the species where relevant. If they ask for “small-space,” provide dimensions. If they ask about compatibility, name the supported model or standard. Useful optimization adds verifiable information rather than vague adjectives.

Put details in the right fields

Use structured attributes for color, size, material, pattern, age group, gender, condition, identifiers, and product type when those fields apply. Keep the title readable and use the description for context. Structured data helps prevent one overloaded text field from carrying the whole product story.

Improve the website at the same time

If the feed says a product has an important feature but the landing page never explains it, the customer will hesitate. Add the same verified details to the visible product page, product schema, specifications, FAQs, and comparison content where useful. The feed and the site should tell one consistent story.

A 60-minute Merchant Center audit

1. 10 minutes: Open Needs attention and record the number of disapproved products, warnings, and account-level issues.
2. 10 minutes: Sort the affected products by issue and export the list for a repeatable work queue.
3. 15 minutes: Compare five high-value products across the store, feed source, and Merchant Center.
4. 10 minutes: Test price, availability, image, canonical URL, and add-to-cart behavior on mobile.
5. 10 minutes: Review shipping and return settings against the published website policies.
6. 5 minutes: Assign an owner and a date for each fix, then schedule a weekly check through the holiday period.

This creates a baseline before anyone changes titles or chases an AI visibility percentage. After corrections sync, allow time for Google to recrawl and re-review the affected products.

What not to do

• Do not request repeated reviews before fixes are live. Failed reviews can lead to cooldown periods.
• Do not edit only inside Merchant Center if the connected store will overwrite those changes.
• Do not invent attributes because a conversational term appears popular.
• Do not use the same generic image for multiple products. The listing should help a shopper identify the exact item.
• Do not treat share of voice as revenue. Pair visibility with clicks, conversions, margin, returns, and real order quality.

Merchant Center AI shopping visibility FAQ

Does every Merchant Center account have AI performance insights?

No. Google currently documents availability for English-language queries in Australia, Canada, India, New Zealand, and the United States. Eligibility and sufficient data can also affect what appears.

Will better product data guarantee placement in AI Mode?

No. Accurate data improves eligibility and understanding, but Google does not guarantee that a product will appear for a particular query or surface.

How quickly will a corrected product return?

Timing depends on the issue. Some website crawl fixes may take up to 48 hours to reflect, while policy or account reviews can take longer. Check the issue details before submitting another review.

Should WooCommerce stores edit the product or the feed plugin?

Usually the best fix is made in the system that owns the incorrect field. That may be the WooCommerce product, a variation, a feed rule, an attribute mapping, or a connected plugin. Trace the value before editing it.

CodeForce can help with business technology systems, WooCommerce and website improvements, Google Ads support, and a clearer WooCommerce checkout experience.

Sources

• Google: Agentic commerce updates and holiday product-feed guidance
• Google Merchant Center Help: AI performance insights
• Google Merchant Center Help: Product and account issues
• Google Merchant Center Help: Product data quality violations
• Google Merchant Center Help: Return policies

Need a second set of eyes on a product feed before holiday traffic ramps up? Book a free CodeForce intro call.