By Steve Merrill, Founder of WRKNG Digital | September 23, 2026
An AI shopping agent runs a quick checklist before it puts you on a shortlist. Pass it and you compete. Fail one and you're out, no matter how good the product is.
Here are the seven checks, in roughly the order an agent runs them.
1. Is it actually in stock?
The agent confirms availability first, because recommending a sold-out item is the fastest way to lose the shopper's trust.
If your stock status isn't machine-readable, the agent can't confirm it and moves on.
2. Can I confirm the price?
An agent won't recommend what it can't price. A price that loads by JavaScript after the page reads as missing.
It also cross-checks the feed against the page. Disagreement is a red flag it resolves by skipping you.
3. Do the facts match across sources?
Feed, page, and structured data need to agree. Contradictions make the agent uncertain, and uncertainty loses to a store that's consistent.
The Agentic Commerce Protocol exists precisely so agents can act on data they trust. Consistency is the price of entry.
4. Can it read the specs?
Specs as text get read. Specs in an image are a wall. An agent matching a detailed request needs the details in a form it can parse.
This is where identical products split. The one written out gets shortlisted.
5. Does it meet the constraints?
Price ceiling, shipping deadline, size, color. The agent filters hard on the shopper's constraints, and it needs your data to check each one.
A blank on a constraint field is treated as a fail, not a maybe.
6. Is there a trust signal?
Structured reviews, ratings, a real brand. These give the agent something to weigh when two products are otherwise tied.
Unmarked reviews barely count. The agent needs a signal it can read, not a wall of testimonials.
7. Can it complete the action?
Increasingly, agents check whether they can actually transact, clean checkout data, no dead ends. A recommendation it can't act on is worth less.
Stores wired for agentic checkout will pull ahead here as the protocols roll out.
Further reading
- Agentic Commerce Protocol (OpenAI + Stripe)
- OpenAI on shopping in ChatGPT
- Schema.org Product reference
Frequently Asked Questions
What do AI shopping agents check before recommending a store?
Roughly in order: is the item in stock, can the price be confirmed, do the facts match across feed and page, can the specs be read, does it meet the shopper's constraints, is there a trust signal like structured reviews, and can the agent complete the purchase. A blank on any of these reads as a fail, not a maybe.
Why does an agent skip a product with great reviews?
If the reviews aren't marked up with schema, the agent can't read them as a trust signal, they're just a wall of text. Add review and rating markup so the agent has something it can weigh when two products are otherwise tied.
What is the Agentic Commerce Protocol?
It's a standard from OpenAI and Stripe that lets AI agents act on structured product and checkout data they can trust, so an agent can recommend and even complete a purchase without guessing. Stores that expose clean, consistent data are the ones agents can actually use.
Want to know if AI assistants can actually find and recommend your store? Get a free AI-visibility read on your Shopify store at WRKNG Digital. We show you exactly what ChatGPT, Perplexity, and Google AI see when they look at your products.
