Your product data is the new storefront

Last month, Amazon expanded its “Buy for Me” feature to more brands. Google’s shopping graph now pulls product details from structured data across the web. And at least three startups I’ve talked to recently are building AI agents that place orders on behalf of procurement teams.

The pattern is clear. The next wave of buyers won’t be human. They’ll be software that reads your catalog, compares it against competitors, and places an order without ever loading your homepage.

If your product data is a mess, you’re invisible to these agents.

The problem with most product catalogs

I’ve looked at a lot of B2B catalogs over the years. The common issues haven’t changed much. Incomplete descriptions. Missing specs. Inconsistent units of measure. Three different ways to spell the same thing across categories.

That was tolerable when a human buyer could call their sales rep and sort it out. An AI agent won’t pick up the phone. It will move on to the next supplier whose data is clean.

According to a 2025 survey by Salsify, 87% of B2B buyers say product information quality directly affects their purchasing decisions. That number is only going up as agents enter the mix.

What “ready” actually means

You don’t need a massive data overhaul. You need to get the fundamentals right.

Structured schema on every product page. If your product pages don’t include Schema.org markup or JSON-LD, most AI crawlers will struggle to parse them. This is table stakes. Google’s own documentation says structured data helps both search engines and AI systems understand your products.

Consistent attributes across your catalog. If “weight” is listed in pounds for half your SKUs and kilograms for the other half, an agent comparing products will either skip you or get it wrong. Pick one system and stick with it.

Complete specs, not marketing copy. Agents don’t care about your “premium quality” language. They care about dimensions, materials, lead times, minimum order quantities, and compatibility. If you sell industrial parts, include the part number cross-references. If you sell chemicals, include the safety data sheets.

Accurate availability and pricing. This one is obvious but hard. If your site shows “in stock” and the agent places an order, you’d better be able to fulfill it. Agents that get burned by bad availability data will learn to avoid you. Yes, agents learn.

The B2B wrinkle

B2B adds a layer of complexity that consumer catalogs don’t have. Tiered pricing. Customer-specific contracts. Volume discounts. Freight classes. Lead times that vary by warehouse.

Most AI agents right now are built for the consumer side of things. But the B2B agent startups I’ve seen are all solving the same core problem: how do you represent complex purchasing rules in a way an agent can read and act on?

The answer starts with your data. If your contract pricing lives in a PDF somewhere and your catalog lives in your PIM and your lead times live in your ERP, no agent can piece that together. You need a single source of truth that an external system can query.

This doesn’t mean you need a massive integration project. APIs over your existing systems work fine. But the data has to be clean, current, and consistent.

A quick test you can run today

Open your site’s product page and view the source. Search for “schema” or “Product.” If you find structured JSON-LD with name, price, availability, and description, you’re in decent shape.

If you don’t find it, that’s your first project.

Next, pick 20 random SKUs and check whether the same attributes appear in the same format for each one. If you see wild variation, you’ve found your second project.

Then try asking ChatGPT or Claude to compare two of your product pages. See what the model can extract. That’s a rough proxy for how an AI buying agent will see your catalog.

Why this matters right now

The AI buying agent space is early. Most of the tools are pilot programs or narrow use cases. But the infrastructure layer is being built now. Google, Amazon, and Microsoft are all investing in product data standards because they know agents need clean, structured information to function.

The merchants who invest in data quality in 2026 will have a real advantage when agent-driven purchasing hits volume. The ones who wait will be playing catch-up.

I’ve seen this movie before with mobile commerce. The merchants who had responsive sites and clean mobile checkouts in 2014 cleaned up when mobile traffic crossed 50%. The ones who waited spent two years fixing technical debt while their competitors took orders.

This is the same cycle, just faster.

What I’d do

Start with the highest-revenue 100 SKUs in your catalog. Get the schema right. Standardize the attributes. Make sure pricing and availability are accurate. Then expand from there.

Don’t hire a consultancy to do a six-month data governance initiative. Just fix the most important products and build from there.

What does your product data look like right now? Have you started thinking about how AI agents will interact with your catalog?