Why You Show Up in ChatGPT but Not Google AI Mode

A pattern comes up in almost every conversation we have with ecommerce leaders. Ask ChatGPT for the best option in their category and the brand appears. Ask Google AI Mode the same question and it does not, or it appears for one phrasing and vanishes for the next. The two systems are not reading the same thing.

ChatGPT builds its answer largely from web pages. Google AI Mode builds shopping answers from the product data Google already holds about you, and that is where the gap usually sits.

Where Google AI Mode gets its product data

Google's conversational shopping in AI Mode is powered by the Shopping Graph, which Google described in November 2025 as more than 50 billion product listings, 2 billion of which are updated every hour. That update rate is what lets AI Mode show current price and stock inside the answer. In April 2026 Google extended the same approach to India, rolling out what it called its biggest upgrade to shopping in AI Mode there alongside shopping in the Gemini app and Circle to Search.

The Shopping Graph is fed mainly by Merchant Center. Google's own guidance on appearing in AI responses says that using Merchant Center feeds helps products be visible in AI responses as well as in other Search results. Its checklist for AI features in Search adds two items that matter for retailers: structured data on the page should match the visible text, and Merchant Center information should be up to date.

This explains the ChatGPT versus Google AI Mode split. A product with a thorough page but a disapproved feed item, a missing GTIN or a guessed category can still be read by ChatGPT and still be invisible as a product in AI Mode. The page is not the problem. The record Google holds is.

Why a product is missing from AI Mode

When a brand asks us why its products do not appear, the cause is almost always one of four things, and all four live in product data rather than in page copy.

• Eligibility. A product that is disapproved or limited in Merchant Center is out before any ranking happens. The product data specification lists the required attributes, and the Issue Details page in Merchant Center shows what is failing per product.
• Guessed categories and attributes. Where a feed leaves category, color, size, material or gender blank, Google fills what it can from the page and the image. Those guesses are often wrong, and a shopper's question about a waterproof boot for wide feet cannot match a product Google has filed under the wrong category with no width attribute.
• Stale price or stock. AI Mode shows price and availability inside the answer. If the feed and the page disagree, or the feed updates once a day while prices move, the product can show the wrong number or be left out of the comparison.
• Thin facts. A controlled experiment published by O'Reilly in April 2026 gave an AI shopping agent a choice between a cheaper jacket described with marketing copy and a dearer one described only with a structured water resistance value. In ten runs it chose the structured, more expensive product every time. Facts the assistant can resolve beat adjectives it cannot.

Google has also added optional conversational attributes to the Merchant Center specification, including question and answer pairs, document links and variant options, which it says help AI systems understand product nuances across AI Mode and other surfaces. They do not affect approval status, but they give the answer more to work with.

Why the visits are worth the effort

The traffic AI surfaces do send behaves like late-stage shoppers. Adobe data reported by Digital Commerce 360 in October 2025 showed visitors from generative AI services spent 32 percent more time on site, viewed 10 percent more pages and bounced 27 percent less than visitors from traditional search. Being absent from the answer costs you the shoppers who had already decided what they wanted.

Five checks on your top products

Start with the products that carry revenue, not the whole catalog. For most brands that is a few hundred items.

1. Open Merchant Center diagnostics for those products. Clear disapprovals, missing identifiers and price or availability mismatches first. Fix them at the source so they stay fixed on the next feed run.
2. Confirm the category and the attributes Google reads. Set the product category yourself rather than leaving it for Google to infer, and fill color, size, material, gender, age group and the product detail and highlight fields with the facts shoppers ask about.
3. Make feed and page agree. Price, sale price, availability and variant data should match between the feed, the page and the structured data on the page. Google asks for that match explicitly in its AI features guidance.
4. Write the missing facts on the page. If a shopper's question is about fit, compatibility, materials or what is in the box, the answer should be on the product page in plain words, not only in a PDF or a support article. The same facts feed the description and product detail attributes.
5. Use the AI performance report. Merchant Center's AI performance insights report shows your share of voice against named competitors for conversational shopping queries on AI Mode and AI Overviews, in the US and a handful of other countries. It is the closest thing to a first-party answer to the question of whether you show up.

How to test it yourself and record before and after

Write ten to twenty questions real customers ask, in their words and situations, and ask them in Google AI Mode and in ChatGPT on the same day each week. Record three things for each: did you appear, did you appear as a product with price and image or only as a link, and which product was named. Because the answers vary between runs, judge the share of appearances over several weeks rather than any single result.

Keep the baseline, ship the data fixes, then run the same set again. Report appearances, AI-referred sessions and orders as three separate numbers. For the analytics side, including why Google AI Mode traffic lands in Organic Search rather than the AI Assistant channel, see how to measure AI referral traffic in GA4 and Shopify.

Frequently asked questions

Why do we show up in ChatGPT but not Google AI Mode?

Because the two draw on different inputs. ChatGPT reads your web pages and third-party coverage. Google AI Mode builds shopping answers from the Shopping Graph, which is fed mainly by your Merchant Center feed. A product that is disapproved, mis-categorized or thinly attributed in the feed can be readable on the page and still absent from AI Mode.

Why does one query return us and the next does not?

AI answers are rebuilt each time, and small changes in phrasing change which products match. Products with complete, accurate attributes match more phrasings. Track appearance rate across a set of questions over time instead of judging from a single search.

Do we need a new feed or a new protocol to appear in AI Mode?

No. Google's guidance is that no new machine-readable files or markup are needed for AI features. The work is the Merchant Center feed you already run, the structured data already on your pages, and the facts in your product content, kept complete, consistent and current.

This is the work Paz does for lean ecommerce teams. A Paz expert reviews your top products across the Merchant Center feed, the product pages and the answers Google AI Mode and ChatGPT give for your shoppers' questions, then proposes the fixes with the evidence behind each one. You approve, and the platform publishes the changes and tracks how your products appear before and after. Talk to us.