Product-level AI visibility is how often AI assistants show or name one specific product, not only the brand that makes it. It is also called SKU visibility. A SKU (stock keeping unit) is one sellable item, such as a 50 ml sunscreen. You measure it product by product, over many answers.
It matters because buyers ask AI for a product. Take a question like "best sunscreen under ₹500 for oily skin". The answer can name one product in its text. Where shopping in ChatGPT is live, ChatGPT can also show product cards, each with an image, product details and links to stores (OpenAI Help Center, checked 7 Oct 2026). A brand can be named often while its bestseller is never named or shown.
This guide covers how ChatGPT shows products according to OpenAI and what you can measure for each product. It also covers what helps, an audit you can run by hand, and how Depra AI measures it.
What is product-level AI visibility?
Product-level AI visibility is the share of AI answers to your buying questions that show or name one of your products. Brand-level visibility counts the answers that name your brand at all.
The written answer can name a product. Or the engine can show it as a product card. A card is a small panel for one product, with a link to a store that sells it.
| Question | Brand-level visibility | Product-level (SKU) visibility |
|---|---|---|
| What is counted | Answers that name your brand | Answers that show or name one product |
| Where it shows | In the written answer | In the written answer and on product cards |
| What you learn | Whether you are on the shortlist | Which product stands for you, and which store the buyer is sent to |
| What you work on | Coverage on the sites AI cites | Product pages, product data and listings |
Why do AI assistants recommend products and not only brands?
Because buyers ask for products, and ChatGPT is built to answer with them. OpenAI's help page says ChatGPT can show product options when a question suggests shopping intent. The options come with images, product details and links to sites where the shopper can learn more or buy (OpenAI Help Center, checked 7 Oct 2026).
The same page says a product appears when ChatGPT judges it relevant to what the user wants. If the user names a budget, ChatGPT focuses more on price.
So an AI product recommendation is specific. It is a named product, and ChatGPT can show it with a price and the stores that sell it.
How does ChatGPT show products?
ChatGPT can show products as cards in a carousel, a row of cards, inside its answer (OpenAI Help Center, checked 7 Oct 2026). OpenAI says shopping in ChatGPT is live for users in the US today and will expand to more regions (OpenAI merchants page, checked 7 Oct 2026). If you ask questions for India, you may see few or no products until ChatGPT shows products there. The written answer and its sources do not depend on this.
OpenAI's help page, Shopping with ChatGPT Search, describes how the cards work. Six points from it matter to a brand (OpenAI Help Center, checked 7 Oct 2026).
- How products are picked. ChatGPT considers what the user asked and context such as memory. It also considers structured metadata from first-party and third-party providers, such as price and product description, and other third-party content. Structured metadata means product facts held in fixed fields. Two more inputs are the model's own answer before it looks at new search results, and OpenAI's safety and product policies. Not every available product is shown.
- Titles and descriptions. ChatGPT may write simplified product titles and descriptions, because merchants often title the same product differently.
- Labels. Some product images carry labels such as "Budget-friendly" or "Most popular". ChatGPT generates them, and OpenAI says they are not verified statements.
- Reviews. ChatGPT may show a summary of reviews. The model writes it from reviews on public websites, and OpenAI does not verify reviews or ratings.
- Stores. When a user clicks a product, ChatGPT may list the merchants that sell it. OpenAI says the list is ranked on factors such as availability, price and quality. Another factor is whether the merchant is the maker or primary seller.
- Not ads. OpenAI says these product results are selected independently by ChatGPT. They are not ads and are not influenced by OpenAI partnerships. Ads are separate.
Shopping research is a separate feature
ChatGPT also has a feature called shopping research. You start it from products already in the chat, and it asks follow-up questions. After a few minutes it returns a buyer's guide. The guide has a short list of top picks, the trade-offs of each and links to merchants (OpenAI Help Center, checked 7 Oct 2026). Our ChatGPT shopping guide covers both features.
What can a brand measure for each product?
For each product you can measure four things: whether it was shown, for which questions, in what position, and through which store. Two more come from the written answer and its sources.
AI answers change from one run to the next, so one answer proves little. Count over many answers and report a share, such as shown in 6 of 20 answers.
The same product can be shown with your own store on the card or with a marketplace. Our guide to marketplace visibility in AI answers covers that side.
| Measure | What it tells you | Where to look |
|---|---|---|
| Shown or not | Whether the product appeared as a card | Product cards in ChatGPT |
| For which questions | Which buying questions trigger products at all, and which trigger yours | The questions you ask or track |
| Position | Where the product sits among the products shown | The order of the cards |
| Which store | The merchant the card points to: your own store, a marketplace or a reseller | The store named on the card |
| Named in the text | Whether the written answer names the product, a rival's product or neither | The answer text, on any engine |
| Sources | Which websites the answer cites | The source links with the answer |
How do you get products recommended by ChatGPT?
Let OpenAI's search crawler in, keep product data accurate and visible, keep price and stock current, and earn real reviews. No source promises a placement. Each point below names its source.
- Let the search crawler in. OAI-SearchBot is OpenAI's search crawler, the bot that reads pages for ChatGPT search. OpenAI says sites that opt out of it are not shown in ChatGPT search answers, though they can still appear as navigational links (OpenAI's crawler documentation, checked 7 Oct 2026). Check your robots.txt, the file that tells bots which pages they may read.
- Keep product data structured and visible. OpenAI's shopping help page says ChatGPT considers structured metadata such as price and product description (OpenAI Help Center, checked 7 Oct 2026). Structured data is the labelled facts in a page's code. Google says it should match the visible text, and that its AI features need no special markup (Google Search Central, checked 7 Oct 2026).
- Know what a product feed asks for. A product feed is a file of product data that a merchant sends to a platform. OpenAI's feed reference has nine required fields. They are an ID, a title, a description, a product page link, the brand, the seller name, an image, availability and price (OpenAI developer documentation, checked 7 Oct 2026).
- Keep price and stock current. The feed reference tells merchants to update price and availability when they change (OpenAI developer documentation, checked 7 Oct 2026). Availability and price are also two of the factors OpenAI names for ranking the merchants on a product.
- Earn real reviews on public sites. OpenAI says ChatGPT's review summaries are written from reviews on public websites (OpenAI Help Center, checked 7 Oct 2026). Its feed reference also has optional fields for review count and star rating.
What a brand in India should know about feeds
OpenAI's documentation says feed onboarding is currently available to approved partners (OpenAI developer documentation, checked 7 Oct 2026). The feed reference says the standard upload currently targets the US, and that eligibility does not guarantee display (OpenAI developer documentation, checked 7 Oct 2026).
Shopify stores have their own route. OpenAI says product data from merchants on Shopify is already integrated into ChatGPT through Shopify Catalog, with no additional work for the merchant (OpenAI Help Center, checked 7 Oct 2026). Shopify Catalog is how Shopify shares a store's product data with channels such as ChatGPT. Shopify says the store must sell to customers in the United States, though it can be based elsewhere, and its products must be eligible for Shopify Catalog (Shopify Help Center, checked 7 Oct 2026).
A feed is not the only way in. OpenAI says a feed is not required if ChatGPT already crawls your site, though a feed gives more control over how products appear (OpenAI merchants page, checked 7 Oct 2026). So for a brand that sells only in India, the public route is the part you control today. That means product pages a crawler can read, the same product facts everywhere, and reviews people can see.
How do you audit your top products by hand?
Use a spreadsheet and nine steps. Start a new chat for every question, signed out or with memory off. OpenAI says ChatGPT also considers context such as memory and custom instructions when it picks products (OpenAI Help Center, checked 7 Oct 2026).
For a ready-made sheet, see how to track brand mentions in AI answers.
- Step 1. Pick the products. Choose 5 to 10 products that matter most, such as bestsellers and new launches.
- Step 2. Write the questions. For each product, write three buying questions that do not name your brand: one by use, one by price and one comparison. For a sunscreen: "best sunscreen for oily skin", "best sunscreen under ₹500" and "gel or cream sunscreen for humid weather".
- Step 3. Ask ChatGPT several times. Ask each question with search on, at least three times, on different days.
- Step 4. Record the cards. Note whether product cards appeared and whether your product was one of them. Note its position, the store named on the card and the rival products shown. If no cards appear, write that down and move on. Steps 6 and 7 still apply.
- Step 5. Open the store list. Click your product, or a rival's, and write down the merchants listed and their order.
- Step 6. Read the text and the sources. Note whether the written answer names your product. Open every source link. Those pages are where to get your product covered.
- Step 7. Repeat on the other engines. Ask the same questions in Gemini, Perplexity and Google, and record whether the answer names your product.
- Step 8. Check your own product page. Confirm that robots.txt allows OAI-SearchBot and that price and availability are in the visible text. Check that the product name matches your marketplace listings.
- Step 9. Repeat every month. Use the same questions each time, so the months can be compared.
How does Depra AI track products in AI answers?
Depra AI records the products ChatGPT showed in each answer it stores for your tracked questions. It calls an answer with products a shopping answer.
It records products whenever ChatGPT shows them. OpenAI says shopping in ChatGPT is live for users in the US today (OpenAI merchants page, checked 7 Oct 2026). So a brand that tracks questions for India may see few or no products in its Shopping report until ChatGPT shows products there.
When your answers do show products, the Shopping report answers four questions.
- Who is shown: brands, ranked by the share of shopping answers that show their products.
- How much: for each brand, the number of shopping answers and that share as a percentage. Also the average position of its products, and the number of its products (SKUs) that appeared.
- Through which stores: the top merchants, meaning the stores the shown products are sold by.
- For which questions: the prompts that trigger products, and the top products.
Limits to know
The percentages are counted over shopping answers only, not over every answer. Only ChatGPT returns products. Gemini, Perplexity and Google AI Overviews, the AI summaries above some Google results, are tracked for brand mentions and sources, not for product cards.
For those engines, track a product prompt: a buying question written for one product. Depra AI then records, answer by answer, whether it names that product, a rival's product, or neither.
When you set up a project for your own store, Depra AI reads the store's product list, so it knows which products are yours. It never changes a store. The store catalogue docs explain how.
Where to read more
See the product visibility feature page and the Shopping docs. Marketplace intelligence covers the stores, and citation gap analysis covers the cited sites. Store owners can start at Depra AI for Shopify stores. Plans are on Depra AI pricing.