How to get your products into Businesses in ChatGPT 

What OpenAI announced 

Businesses in ChatGPT has four parts:

• Business Hub. Verified businesses can claim their profile, update their details, and add self-service product feeds with pricing and availability.
• Insights. The Hub reports mentions, website clicks, and “discovery topics,” the common themes across conversations where your site is cited. Insights are aggregated and appear only above privacy thresholds.
• Business Agents. These answer shoppers’ follow-up questions and help compare products. They’re clearly labeled, separate from ChatGPT’s own answers, and currently available to eligible U.S. ecommerce and online service businesses.
• Sponsorship. Business Agents can be sponsored and used as a ChatGPT Ads destination.

OpenAI also plans tools for organizations with more complex needs, a signal manufacturers and enterprise brands should watch.

Why structured product data will decide who wins

Search rewarded brands with the best pages. AI shopping rewards brands with the best data.

When a shopper asks ChatGPT which of two drills has more torque, or for a waterproof trail shoe under $100, there’s no list of links to scan. ChatGPT builds an answer from what it can match: attributes, specs, variants, and prices. Products with complete, structured data can be matched and compared. Products described only in marketing copy are easy to pass over.

OpenAI’s own developer documentation says feed quality improves discovery relevance, and every part of Businesses in ChatGPT, from product feeds to Business Agents, runs on product data.

That shifts the advantage to brands that treat product data as a strategic asset: structured, enriched, and governed once in a PIM, then published everywhere. In the next era of ecommerce, the winners won’t just have the best products. They’ll have the best-described ones.same place.

How your product data reaches ChatGPT

There are two ways in: through Shopify, or through a feed you send directly. For Shopify merchants, the path starts in the PIM.

• Step 1: Inriver to Shopify. The certified Inriver Shopify Adapter syndicates products, variants, metafields, categories, and images to your store, on a schedule or on demand, across multiple stores.
• Step 2: Shopify to Shopify Catalog. Catalog structures each product’s title, description, options, images, price, availability, and key attributes for AI agents, and keeps them updated continuously.
• Step 3: Shopify Catalog to ChatGPT. For eligible stores, discovery is on by default, and no separate feed is required. You choose which AI channels can use your Catalog data under Sales Channels > Agentic in Shopify admin.

To be clear: Inriver doesn’t send data to ChatGPT directly, Shopify and OpenAI decide eligibility, and price and inventory usually come from Shopify or your ERP. What the PIM controls is everything that describes the product: titles, descriptions, attributes, variants, and images.

Not on Shopify? Structured product data is your way in

Shopify isn’t the only route. In Business Hub, verified businesses can add self-service product feeds directly, and OpenAI’s message across the launch is consistent: give ChatGPT structured, accurate product data.

OpenAI’s developer documentation says product feeds supply the structured catalog data that helps ChatGPT surface the right products, and that feed quality improves discovery relevance. Feeds can be delivered by SFTP, file upload, or hosted URL. The product feed spec reads like a PIM requirements list:

• Nine required fields per item: ID, title, description, product URL, brand, seller name, main image, price, and availability.
• Variants as separate rows, grouped under a parent ID, with their distinguishing options spelled out.
• Optional attributes such as category, material, color, size, dimensions, weight, GTIN, and manufacturer part number.

Two details matter for data teams. Units aren’t inferred, so measurements must be converted and labeled before upload. And if a product’s attributes disagree with its variant options, OpenAI won’t sort out which one is right.

When your product data is already structured and governed in Inriver, a ChatGPT feed starts from data you trust, not a spreadsheet cleanup. The same record can serve Shopify, a direct feed, retailers, and marketplaces.

Business Agents are only as smart as your product data

Getting into ChatGPT is step one. Holding up when a shopper starts asking questions is harder.

OpenAI’s own example: a shopper compares two products and asks which features fit their needs. Answering well takes attributes that are complete, consistent, and comparable across your catalog. Four gaps get in the way:

• Missing specs. If the dimension, material, or rating a shopper asks about isn’t there, AI has nothing to match.
• Inconsistent units. Watts next to kilowatts, or inches next to centimeters, make comparison unreliable.
• Vague descriptions. Marketing language without concrete facts gives AI little to work with.
• Undifferentiated variants. Without the attributes that set variants apart, AI can’t say which one fits.

Each gap gives AI less to match against a buyer’s question, and makes it more likely a competitor’s better-described product gets recommended instead.

This is what a PIM is built for. In Inriver, teams enrich attributes and connect products to related content such as compatibility, components, and use cases, then publish it to Shopify as metafields and metaobjects.

Your first real AEO dashboard

Measuring answer engine optimization (AEO) has mostly meant indirect evidence like sample prompts and referral traffic. Business Hub adds first-party data from ChatGPT itself: mentions, website clicks, and discovery topics.

Treat these as leading indicators of visibility, not pipeline results, and connect them to site engagement, conversions, and revenue. Discovery topics are the most useful for product teams, because they show the themes you’re cited for. Compare them with the themes you want to own. If buyers in your category care about durability or compatibility and you’re not cited for those, the gap usually points to attributes or content to fix in the PIM.

One limitation: insights appear only above privacy thresholds, so lower-volume brands may see little data at first. Build the review habit anyway.

One source of truth across every AI channel

OpenAI calls user trust its North Star. Product data deserves the same standard: when your website and a ChatGPT answer show different specs, both lose credibility.

ChatGPT isn’t the only destination, either. Shopify Catalog also feeds Copilot, the Shop app, and other AI agents and apps, alongside your retailers and your own site.

The way to keep them aligned is to enrich once in the PIM and publish everywhere. Then your website, ChatGPT profile, retailer listings, and AI agents draw from the same data and stop contradicting each other.

Readiness checklist: 8 steps to prepare your product data for ChatGPT

Use this list to move from “our products are in ChatGPT” to “our products compare well in ChatGPT.” 

• Choose your path. Not on Shopify? Map your catalog to OpenAI’s product feed spec. On Shopify? Confirm your store’s Catalog eligibility and review which AI channels can use your data under Sales Channels > Agentic.
• Claim your business in Business Hub. Shopify merchants can connect their store to speed up verification.
• Audit attribute completeness. Start with your top-revenue products and the products customers ask the most questions about.
• Standardize for comparison. Align units, naming conventions, and variant attributes so products can be compared side by side.
• Map enriched attributes to Shopify. Use the Inriver adapter to map PIM attributes to Shopify metafields.
• Add related content. Connect products to compatibility details, use cases, and FAQs.
• Review monthly. Check Business Hub mentions and discovery topics, and route gaps back to the PIM as enrichment tasks.
• Assign owners. Decide who manages the business profile, the product feed, and the Business Agent.

Make your product data count

Your products may already be in ChatGPT. The real question is whether your data gives AI enough to recommend them. Talk to us about the Inriver Shopify Adapter and how enriched PIM data reaches Shopify and every AI channel beyond it. 

Want to go deeper first? Read PIM for Shopify: The complete integration guide.