How to Get Products Recommended by ChatGPT and AI Search

To get your products recommended by ChatGPT and other AI search tools, give each assistant clean structured product data it can trust: a product feed (OpenAI's feed for ChatGPT, Google Merchant Center for Google Search and Gemini), Product markup on every product page that matches the visible page, crawlable pages, and real reviews. Assistants recommend what they can parse, price and verify.

If you want ideas like these working inside your own business, see our AI development and automation service.

If your pages are not being crawled at all, start with our guide on how ChatGPT, Perplexity and AI Overviews choose what to cite, because no feed fixes a blocked site.

How do AI assistants decide which products to recommend?

AI assistants decide which products to recommend by matching the shopper's request against structured product data (price, availability, attributes, reviews) and against what the open web says about those products. The assistant then filters for the shopper's stated constraints, such as budget or size.

Two inputs feed that decision:

1. Data you submit directly, such as a product feed to OpenAI or Google Merchant Center. This supplies price, stock and identifiers.

2. Data found on the web: your product pages, their markup, and reviews and comparisons elsewhere. This shapes whether the product looks any good.

OpenAI says its required feed fields "ensure correct display of price and availability", while recommended attributes "like rich media, reviews, and performance signals" "improve ranking, relevance, and user trust" (OpenAI Agentic Commerce key concepts). The required fields get you in. The extra fields decide whether you get picked.

How do you get your products into ChatGPT shopping results?

You get products into ChatGPT shopping results by submitting a product feed to OpenAI and keeping your product pages open to OpenAI's search crawler. As of November 2026, OpenAI says feed onboarding "is currently available to approved partners", and merchants apply through the form on chatgpt.com/merchants (OpenAI commerce getting started).

The feed itself is specific. Per the OpenAI product feed spec:

• Nine fields are required: item_id, title, description, url, brand, seller_name, image_url, availability and price.

• availability must be one of in_stock, out_of_stock, pre_order, backorder or unknown. A missing or unrecognized value rejects the row.

• Titles should be at most 150 characters and descriptions at most 5,000.

• gtin is optional but, when present, must be exactly 8, 12, 13 or 14 digits with a valid check digit.

• Reviews go in as review_count together with star_rating.

• Search eligibility is on by default (is_eligible_search), checkout is off by default (is_eligible_checkout), and ads eligibility is a separate flag.

Two lines in the spec matter most. First, "Eligibility does not guarantee display." Second, the standard upload "currently targets the US", so brands selling elsewhere should confirm their countries with OpenAI before building anything. Files go over SFTP as a full snapshot, at least daily.

Instant Checkout is a separate step. Turning on is_eligible_checkout does not complete checkout onboarding, which also needs the Agentic Checkout and Delegated Payment specs. Our comparison of UCP, ACP, AP2 and MCP explains which checkout protocol a store needs.

How do products show up in Google AI Mode and AI Overviews?

Products show up in Google AI Mode and AI Overviews through the same systems that power normal Google Search and Shopping: indexed product pages and Google Merchant Center data. Google states that "there are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary" (Google Search Central, AI features and your website).

The same page lists what still matters: crawling allowed in robots.txt and by your CDN, important content in text form, structured data that matches the visible text on the page, and "checking that your Merchant Center and Business Profile information is up to date". Google also says you need no new AI text files or special schema, which rules out a lot of paid "AI optimization" extras.

Merchant Center is the part small brands skip. According to Google's free listings help page, products can appear at no cost on Google Search, Maps, Gemini, YouTube, the Shopping tab, Google Images and Google Lens, and new Merchant Center accounts are opted in to free listings by default. Required feed attributes include id, title, link, image_link, price, description and availability. gtin is required whenever the manufacturer assigned one, and shipping settings are required in 30 countries, including the US, UK and Australia.

Google can also show products it finds by crawling, even without markup or a feed, but feed data is explicit while crawled data is a guess.

How does Perplexity choose products to recommend?

Perplexity chooses products largely from merchant data it receives through platform integrations, and it says its shopping product cards "aren't sponsored" (Perplexity, Shop like a Pro, November 2024). Shopping discovery is "powered by platform integrations including Shopify", which at launch covered Shopify businesses that sell and ship to the US.

In that November 2024 launch, Perplexity also announced a merchant program. It describes the program as aimed at large retailers, says it "is free for merchants", and lists "increased chances of being a recommended product" among the benefits, along with payment integrations, free API access and a trends dashboard. Smaller D2C brands are not shut out, because the Shopify integration covered Shopify stores that sell and ship to the US at launch, and clean pages with Product markup are likely to help any assistant that reads the web.

What product data makes AI assistants pick your product over a competitor?

AI assistants pick the product whose data answers the shopper's exact question, so the winning product data is specific, consistent and verifiable. Shoppers ask for "an insulated bottle under $40 that fits a car cup holder", and an assistant can only recommend you if your data states those facts in words it can match.

The fields that most often decide it:

• Attributes in plain words. Size, material, compatibility, dimensions and who the product is for, written into the description and the title, not only shown in an image.

• Stable identifiers. A valid GTIN or MPN can help an assistant match your listing with reviews and mentions elsewhere, and Google requires a GTIN whenever the manufacturer assigned one.

• Live price and stock. OpenAI rejects rows with bad availability values, and Google wants structured data that matches the visible page.

• Reviews. Review counts and ratings are optional fields in OpenAI's feed, and OpenAI says reviews improve ranking and user trust.

• Return and shipping terms. Shoppers ask about delivery cost and returns, so state both in text on the page and in the feed.

• Variants done properly. One parent product with a separate entry per purchasable size or color, each with its own price and stock.

This is mostly data hygiene. Our piece on what D2C brands need to fix before buying AI tools makes the same point about operations: messy catalog data limits every AI system built on top of it.

A worked example: one product, three assistants (hypothetical)

This example is hypothetical. A D2C brand sells a 600 ml insulated steel bottle for $34 in the US. Its page shows the price, a photo and brand story; the diameter only appears in an image. There is no GTIN in the markup, no Merchant Center account and no OpenAI feed.

A shopper asks ChatGPT for "an insulated bottle under $40 that fits a car cup holder". The assistant can see the price if it crawls the page, but it cannot confirm the bottle fits a cup holder, so a competitor whose description says "7.3 cm base fits most car cup holders" wins the slot.

The fix, in order:

1. Write the diameter, capacity, insulation hours and "fits most car cup holders" into the description and Product markup, in text.

2. Add the GTIN and brand to the markup and keep price and availability in the markup identical to the page.

3. Open a Merchant Center account, submit the feed and leave free listings on. This puts products on Google Search, Shopping, Gemini, Images and Lens, and Google says current Merchant Center data also matters for AI Mode and AI Overviews.

4. Apply for OpenAI feed onboarding and, if accepted, push a daily full snapshot with review_count and star_rating filled in.

5. Ask recent buyers for reviews that mention the use case, since that is the language assistants match.

What tools help D2C brands show up when shoppers ask AI for product suggestions?

The tools that help D2C brands show up in AI product suggestions are mostly the ones you already have: Google Merchant Center (free listings cost nothing), OpenAI's merchant feed onboarding (by application), your store platform's built in feeds and Product markup, and a crawler check. Paid feed software helps with thousands of SKUs, but it only formats data you already need to have right.

A practical stack:

• Google Merchant Center for Google Search, Shopping, Gemini, Images and Lens, with current data that Google says also matters for AI Mode and AI Overviews.

• OpenAI merchant feed for ChatGPT shopping, through the approved partner program.

• Your store platform's integrations. Perplexity names Shopify as a discovery integration, so check which AI channels your platform already exposes before building anything custom.

• A readiness check. Our free agentic commerce checker tests whether AI shopping agents can read your store, and the GEO checker tests whether AI engines can understand and quote your pages.

• A robots.txt review. Our post on how merchants block AI shopping agents by accident shows the ten minute check.

Syncing product data across a storefront, feeds and a support agent is the kind of build AISmith does through its AI engineering services. For a view of which platforms handle which jobs, see our comparison of AI automation platforms for D2C brands.

Is AI search taking traffic from D2C brand websites, and how do you measure it?

Yes, AI search takes some traffic from D2C brand websites, because assistants can answer a shopping question without sending a click. You measure the effect by tracking where your products appear in AI answers as well as how many visits arrive.

Three checks to run each month:

1. Search Console. Google counts traffic from AI Overviews and AI Mode in the Performance report under the "Web" search type, not as a separate line.

2. Referral traffic. Look for visits from AI assistant domains in your analytics and compare their conversion rate with your search traffic.

3. Prompt testing. Write ten questions your customers really ask, including budget and use case, and run them in each assistant. Record who appears and why, and repeat after every data fix.

What is LLMO for D2C brands?

LLMO, short for large language model optimization, is the practice of making a brand's products and pages easy for AI assistants to find, understand and recommend. For a D2C brand it is the same work as generative engine optimization, applied to a catalog: accurate feeds, product markup, open crawling, specific descriptions and third party reviews. The label is new; the work is product data quality.

Key takeaways

• ChatGPT reads a product feed from approved partners, with nine required fields; as of November 2026 the standard upload targets the US.

• Google says AI Overviews and AI Mode need no special optimization; crawlable pages, accurate Product markup and current Merchant Center data are what matter.

• Perplexity announced (November 2024) Shopify powered discovery plus a free merchant program for large retailers.

• Write shopper attributes into text, add GTINs, keep price and stock live, and test real customer prompts monthly.

Frequently asked questions

Can you pay to be recommended by ChatGPT?

ChatGPT ads exist, but OpenAI's feed treats them separately from organic shopping results. The ads eligibility flag is "independent of search eligibility", and ads feeds follow their own guide. Organic visibility still depends on your product data.

Do I need a separate feed for every AI assistant?

You usually need one clean source of product data and a few outputs from it. Google uses Merchant Center, OpenAI uses its own feed format over SFTP, and Perplexity relies on platform integrations and its merchant program.

Does Product schema markup help with AI product recommendations?

Product markup helps because it gives crawlers price, availability, identifiers and reviews in a predictable form. Google says no special schema is needed for its AI features, but the markup must match the visible text on the page.

How long does it take to show up in AI shopping results?

The timeline depends on the channel. OpenAI feed access depends on partner approval first, and an accepted feed should then be refreshed at least daily. Open web signals such as reviews and mentions build slowly, so start with the data you control.

Do small D2C brands stand a chance against big retailers in AI answers?

Small D2C brands can win AI answers because assistants match specific constraints, not brand size. Clear dimensions, honest reviews and a live price can beat a bigger listing with vague copy.