What 118 Footwear Brands Built for AI Shopping Agents

Of 118 footwear brands I ran through our agent-readiness checker, one runs an MCP server of its own: On Running. 93 could be scored, 25 refused our automated checker, and none of the 93 went above 4 of 9.

It is the same checker and the same nine checks as our eyewear store study. It reads files at public addresses on each store, the things an AI agent can fetch. It does not measure how often AI answers mention a brand. For that, see our UK footwear AI visibility ranking.

Footwear AI agent readiness in numbers: 118 brands checked

We checked the brands between 22 October and 2 November 2026. The 18 checked first were run again on 2 November, so every result below comes from the same version of the checker.

A score is out of 9, or out of fewer when some checks could not be read. A brand that refused our checker gets no score and stays out of every average. Nine brands scored 0.

Which footwear brands built AI agent features of their own

Most points in this study were not built by the brand. On every Shopify store, Shopify supplies an MCP endpoint, a UCP profile and an llms.txt (see Shopify's agentic commerce docs). Take those away, and what the brands did themselves comes to 63 points across 93 brands.

“

“Hyper‑personalization is all about asking what consumers want when they’re visiting my properties and tailoring the experience to their needs.”

54 of the 63 are one thing: a Wikidata entry that names the brand's website. The other nine are below, by name.

On Running has the only brand-hosted MCP server among the footwear stores we read

On Running (on.com), the Swiss brand, scored 4 of 8 readable checks, and every point is its own. It answers an MCP handshake at www.on.com/mcp, serves its own UCP profile, publishes an llms.txt and has a Wikidata entry tied to its site.

On Running uses a custom storefront, so none of this came with a platform. 20 Shopify stores also reached 4 of 9, and for each of them three of the four came with Shopify.

Birkenstock is the one Salesforce footwear store serving UCP

Birkenstock's German store serves a UCP profile, version 2026-04-08. It is the only one of the 19 Salesforce stores in the set that does. The other 18 answered nothing at /.well-known/ucp.

“

“Protocols like UCP turn agentic commerce into something merchants can actually adopt at scale,” said Prakhar Mehrotra, Senior Vice President and Head of AI at PayPal. “Interoperability allows retailers to connect once and reach many environments, while maintaining trust, transparency, and control.”

Six of those 19 scored 0 of 9: Wolverine, HYTEST, Bates, Kodiak, R.M. Williams and Bared Footwear. For what each commerce platform offers agents, see agentic checkout requirements by platform.

Six footwear brands publish an llms.txt of their own

Ariat, Pikolinos and Skechers on Salesforce, Paraboot on WooCommerce, and Camper and On Running on custom storefronts. Five of them wrote the file. Paraboot's is generated by a WordPress plugin the store switched on. What is inside each one differs a lot, so the next section shows them.

llms.txt examples from six footwear brands

An llms.txt is a plain Markdown file at the root of a site that tells AI tools what the site is and where to look. I asked for each file in a browser and read all of them. Four kinds turned up.

Brand-written llms.txt files: Ariat, Skechers, Camper and Pikolinos

Ariat's reads like a merchandising plan. It groups links into core hubs, "strategic targets" and a knowledge base of fit and care guides, and every link carries a tracking parameter:

# Ariat International - AI Context & Navigation Manifest
...
## 1. Core Hubs (Brand Trunk)
- [Men's Cowboy Boots](https://www.ariat.com/mens-cowboy-boots?src=llms.txt) | Advanced technology and classic styling.

That ?src=llms.txt means Ariat can see in its analytics which visits came through the file. Nobody else in the set does this.

Skechers keeps it plain: core pages, a sitemap, the blog, and a separate block for shipping, returns, help and policies. It ends with rules for the reader:

## Crawling Guidance
Please use canonical product and category URLs from the sitemap.
Do not treat cart, checkout, account, search result, or filtered URLs as authoritative content.

Camper's is the most shopper-shaped: categories, a size guide with US/UK/EU conversion, care, materials, US shipping, returns and order tracking. It also lists its own vocabulary in a "Key terms" line (Pelotas, Twins, Beetle, ReCrafted).

Pikolinos sells in 14 country and language stores, and its root file is a router. It points to separate files per market and tells the model what not to assume:

Do not assume that products, prices, promotions, delivery conditions or return policies shown in one market apply to another.

For a brand selling in many countries, that sentence matters more than any list of links.

On Running's llms.txt points to a glossary of its own technology

On Running's file is short. Besides a shoe finder and the catalogue, it links a technology glossary in Markdown and the same glossary as structured data. It is the only file in the set that gives AI tools machine-readable definitions of the brand's own terms.

A plugin-generated llms.txt: Paraboot

Paraboot's file opens with "Generated by Yoast SEO v27.7". Yoast is a WordPress SEO plugin, so nobody at Paraboot chose what went in it.

Its first section is five job ads. The "Produits" section lists repair services, not shoes. An AI reading it learns that Paraboot is hiring before it learns what Paraboot sells.

It still counts as Paraboot's own point, since the store switched the feature on. I doubt the file does Paraboot much good.

The llms.txt Shopify generates for its stores

Every Shopify store gets an llms.txt from the platform (Shopify's developer documentation). Blundstone's is typical, and it is a different kind of file. It says nothing about boots. It is a manual for shopping agents:

# Agent Instructions - Blundstone USA
...
1. Discover the store - GET https://www.blundstone.com/.well-known/ucp returns its UCP profile
...
- Buyer authorization - Place orders and submit payment only with the buyer's authorization.

It tells an agent how to check out, and it ends with Shopify's own pitch ("Start your own store"). That helps an agent that is ready to buy, but nothing in it tells a model what the brand sells, which is what the four brand-written files do.

Where footwear agent readiness points come from, by platform

The same data by platform. "Own" means points the brand earned itself; on Shopify that is the score minus the three Shopify supplies. The last column counts brands that earned none.

Across the 44 brands not on Shopify: 43 points, all their own. "Not identified" means the checker could not read the homepage, so it had nothing to tell the platform from.

One footnote. Native Shoes, on Shopify, refused two of our requests for its files, and the checker credited those two points by platform rule. In a browser its UCP profile is there, so the credit is right, but it was not read by us.

Same company code, same agent readiness score in footwear

Five stores carry the name Caleres in their own page code: Naturalizer, Stuart Weitzman, Sam Edelman, Dr. Scholl's Shoes and Allen Edmonds. All five scored exactly 1 of 8, and the one point is Wikidata each time.

Three stores name Wolverine World Wide as their company on their own pages: Wolverine, HYTEST and Bates. All three run on Salesforce and scored 0.

When one team runs several stores on one template, a brand cannot do better than the template. A fix made once would reach all of them.

Wikidata, the agent readiness point most footwear brands already have

Wikidata is an open database of facts about companies, people and places, and AI tools can read it. Our checker gives a point when a Wikidata entry lists the store's own website as the official website.

That point is 54 of the 63 points brands earned themselves. Eight brands in the study have an entry that does not connect to the store we checked:

• Rothy's, Dansko, Vessi, Tecovas, OOFOS and Sanders have an entry with no official-website statement, so nothing in the structured data ties the entry to their store.
• Loake's entry lists loake.co.uk and Hanwag's lists hanwag.de, while the stores we checked are loake.com and hanwag.com.

For most of them the fix is one statement on an entry that already exists.

25 footwear brands refused our automated checker

Adidas, ASICS, Brooks, Chippewa Boots, Converse, Dr. Martens, ECCO, Grenson, Hanwag, Hey Dude, HOKA, La Sportiva, Mollini, New Balance, Salomon, Sanders, Saucony, Sorel, Teva, Timberland, Topo Athletic, UGG, Vans, Vionic, Vivobarefoot.

Six of them are among the ten best-known brands we started with. Of the other 108, 19 refused.

This is a fact about access from our servers, not a score. Several of these sites load normally in a browser. In a separate test from a home connection, a plain headless browser was refused by 9 of 10 well-known footwear stores.

What it does not say: that AI agents cannot shop there. Agents that buy through feeds and APIs never load the storefront, and agents that drive a browser arrive from their own providers' networks (see how AI agents check out). The move for a store is bot management that lets verified agents through. Removing bot protection invites card testing and scraping.

Grenson is a side example, not counted. It refused our checker, but its homepage source loads Cloudflare's WebMCP bridge, which points browser AI at an MCP server at /api/mcp. That address answers 404 in a browser, so the store switched on an agent feature behind a wall that turns automated visitors away, and the feature points at nothing visible. More on WebMCP in WebMCP for Shopify stores.

How a footwear brand can raise its AI agent readiness score

1. Check your Wikidata entry. If the official website is missing or lists another domain, add the domain your store runs on. It is a single statement, and it is worth a point.
2. On Salesforce, ask whether UCP is on for your store. In this set one Salesforce store of 19 serves a profile.
3. Write your own llms.txt, even on Shopify. Say what you sell, link categories, size guides, shipping and returns, and add rules like Pikolinos does. If a plugin writes it for you, read what it wrote.
4. Check what your firewall does to verified agents. Bot protection should stay, tuned so the agents you want get through.
5. Run your store through our agent-readiness checker and see which of the nine checks you pass.

For the bigger picture of what gets products bought by agents, see how to get products bought by AI shopping agents.

All 118 footwear brands and their agent readiness results

Sorted by score, then by points the brand earned itself. Each name links to that store's report in our directory.

No score (refused our automated checker): Adidas, ASICS, Brooks, Chippewa Boots, Converse, Dr. Martens, ECCO, Grenson, Hanwag, Hey Dude, HOKA, La Sportiva, Mollini, New Balance, Salomon, Sanders, Saucony, Sorel, Teva, Timberland, Topo Athletic, UGG, Vans, Vionic, Vivobarefoot.

The footwear directory shows the same stores as they are checked today, so its numbers move. This post is the snapshot of 2 November 2026.

Footwear agent readiness study method and limits

Dates, checker version and storefronts

Brands were checked between 22 October and 2 November 2026, and the 18 checked first were re-run on 2 November so all 118 rest on one version of the checker. Runs started from Europe. Where a brand serves several countries, the store the checker received is not always the one a US shopper would get.

What counts as a point the brand earned itself

On Shopify, the MCP endpoint, the UCP profile and the llms.txt come from the platform, so they count as Shopify's. On every other platform in this set, nothing came with the platform, so every point is the brand's. The nine checks are explained in our beauty brands study.

Agent readiness and AI visibility measure different things

This study reads files an agent can fetch. Our UK footwear ranking measured how often AI answers name a brand. A brand can do well on one and badly on the other.

The data is published under CC BY 4.0: reuse it freely with credit to GEOlikeaPro and a link to this page.

Want to see where your own store stands? Run it through the checker. It takes about a minute.