AI Visibility Guide
Typical first-scan score
61 / 100
That is the average across 841 retail and e-commerce sites in the Faro Score Index (November 2026), a little above the web average of 59. The average hides the gap that matters: a store can score well on the basics and still show an agent no readable price or returns policy.
Your store now has two kinds of visitors. People, who look at photos and scroll. And AI shopping agents, which read your product data, prices and policies, compare you with other stores, and decide on the person's behalf. Someone asks ChatGPT, Gemini or Perplexity for running shoes under $120 with free returns, and an agent does the shopping. It never sees your hero image. It sees whether your product pages say what the shoe costs, whether it is in stock, and what happens if it does not fit. Stores that answer those questions in a form a machine can read make the shortlist. Stores that hide them behind scripts, pop-ups or vague policy pages get skipped, often without a single human ever visiting.
The agent is becoming the first visitor, and sometimes the only one. Shopping features inside AI assistants let a person go from question to shortlist to checkout without opening ten tabs, and the person often trusts the shortlist as given. That moves the decision earlier: if the agent cannot read you, you are out before the comparison starts, and no ad or design budget wins it back. We see the same shift on our own site. Since August 2026, AI agents have paid Faro directly for scans 15 times, from 10 different wallets, with no person clicking a button. Agents are already acting as buyers. The question for a store is whether they can buy from you.
Product data an agent can read
Product schema with name, price, currency, availability and images lets an agent compare your product with others without guessing. Without it, you rarely make the comparison table.
Prices in the page, not added later by a script
Many agents read the HTML your server sends. If the price only appears after JavaScript runs, the agent sees a product with no price and moves on.
Shipping and returns in plain words
Agents answer questions like 'free returns?' and 'arrives by Friday?'. A short, crawlable policy page with the numbers stated gives them a yes. A vague one gives them 'unknown', and unknown loses.
AI crawler access
If robots.txt or your firewall blocks AI crawlers, your store is invisible to the assistants doing the shopping. Faro checks which bots you let in and which you shut out by accident.
A map for agents (llms.txt)
A short /llms.txt file points agents to your main categories, policies and best sellers, so they spend their time on pages that sell.
Checkout an agent can complete
Agents increasingly hand off to checkout. Guest checkout, clear totals and no forced account creation keep the sale from dying at the last step.
A person asks an AI assistant for running shoes under $120 with free returns that arrive by Friday. The agent checks a dozen stores in seconds. Eight show a readable price. Six state their return window. Four say when an order ships. Those four are compared, and one is recommended with a link to checkout. The other stores never knew they were in the running.
They are AI assistants that shop for people. They search across stores, read product data, prices and policies, compare options and, more and more, hand the person straight to checkout. For a store, they are a new kind of visitor that decides on facts, not on design.
They can if they can read what you sell, what it costs, whether it is in stock and how delivery and returns work, and if checkout does not block them. Run a free Faro scan to see which of those an agent can read on your store today.
It picks the stores whose facts it can confirm: price, stock, delivery time and returns. If two stores sell the same product and only one states its return window clearly, that one usually wins the recommendation.
No. Most of the work is making your existing store readable: product schema, prices in the HTML, plain policy pages, AI crawler access and an llms.txt file. A second site you have to keep in sync tends to go stale and contradict the real one.
Start with the basics agents rely on: Product schema on every product page, crawlable prices and stock, a clear returns page, and robots.txt that lets OpenAI's crawlers in. Then check your store with Faro to find what is still blocking them.
Free scan. No signup. Results in 30 seconds.
Scan your site →More industries