How AI-ready are independent running stores? The fall 2026 benchmark of 282 stores

Short answer: We ran 282 independent US running specialty stores through sqrly Pulse, our AI-readiness assessment. The median store scores 36 out of 100, the highest 73, and no store reached the top band. When we asked four AI assistants to recommend a store in each one's category, 98% of stores weren't named by any of them. And one fact shapes this whole category: 134 stores (48%) run their online store on the same running-store point-of-sale platform, and every one of them scores between 30 and 42.

This is our second benchmark, after ski and snowboard shops. Running stores are a close cousin: specialty retail built on expertise (gait analysis, fit, which shoe for which runner) that shoppers increasingly ask an AI assistant about first. Whether the assistant can read a store's website well enough to name it is the question we measured.

What we did

Between October 2 and 8, 2026, we ran every store on our list through the same sqrly Pulse assessment, on one fixed scoring build and with the same four assistants as the ski benchmark, so the two compare. The list is independent US running specialty stores, single stores and small chains, compiled from industry directories and web search and reviewed by hand. We scanned the site where each store actually sells: when a store's "Shop" link goes to a separate storefront address, we scanned that storefront. We also scanned four national chains as a reference. They're reported separately and are not part of any other number here.

For each store, the assessment reads the public website the way an AI system would (pages, product data, structured data, policies) and asks Claude, ChatGPT, Gemini, and Perplexity about it live, with web search on. The overall score runs from 0 to 100. The full methodology is at the end.

Finding 1: most stores sit low, and they bunch up

Overall scores, 282 stores

Invisible to Agents (0–29)Not Yet Ready (30–49)Partially Ready (50–69)Mostly Ready (70–84)Agent-Ready (85–100)

The median is 36, and the middle half of stores score between 33 and 52. The most striking thing in the chart is the spike: 135 stores scored in the 30s. That isn't a coincidence, as the next finding shows.

(The band names are the assessment's own labels for score ranges. "Invisible to Agents" means a score under 30. It doesn't mean an assistant can't find the store at all.)

Finding 2: one platform sets the range for half the category

Median overall score: one point-of-sale platform vs everyone else

134 of the 282 stores (48%) run their online store on the same point-of-sale platform built for running stores, usually (111 of them) at a "shop." address next to their main website. Every one of those storefronts scored between 30 and 42, with a median of 35. Everyone else has a median of 47.

A range that narrow, across 134 different stores with different staff, catalogs, and marketing, means the platform's storefront template does most of the work, for better and for worse: it gives every store a baseline, and no store on it got past 42. This is a description of what we measured, not a verdict on the platform or the stores. But if your store sells through a shared storefront like this, the most effective fix may be a question to your platform provider rather than a change to your own site: does the storefront publish structured product data, show prices in the page itself, and offer an llms.txt?

We're not naming the platform. The point is the pattern, not the vendor, and it will change as platforms update their storefronts.

Finding 3: assistants almost never name a running store

For each store we asked every assistant a category question with no store named, the way a shopper would ("I'm looking for running shoes and gear — which specific shops or brands should I buy from?", with the category taken from the store's own site), in a fresh conversation with web search on.

• 275 of 282 stores (98%) weren't named by any of the four assistants.
• 2 were named by one assistant, 1 by two, 2 by three, and 2 by all four.

For ski shops, 19% were named by at least one assistant. For running stores it's 2%. This is the noisiest number in the report (one live question per assistant per store), so read it as a pattern. The pattern is clear: when shoppers ask where to buy running gear, assistants point them to shoe brands (Hoka, Asics, Brooks, Saucony, Nike, New Balance) and national retailers (Fleet Feet, Road Runner Sports, Running Warehouse, REI), not to independent running stores.

Finding 4: Shopify stores score the same as in ski; the mix is what differs

Median AI Search Visibility by platform

We compare platforms on AI Search Visibility, which measures what any assistant can learn from a store's public website and is scored the same way for every store. (The overall score adds a platform-shopping channel we can only assess on Shopify and WooCommerce.)

Shopify running stores post a median of 68, almost exactly what Shopify ski shops posted (70). Shopify adds structured product data and an llms.txt file to every store by default, so stores on it start from the same place in any category. The difference between the two benchmarks is mostly the platform mix: about half the ski shops ran on Shopify, but only about a quarter of running stores do.

Finding 5: what the lowest-scoring stores are missing

Among the 201 stores not on Shopify or WooCommerce, the bottom quarter (scores of 30 or less) is missing the basics that the stores above them have:

Off Shopify: what the lowest-scoring quarter is missing (201 stores)

Top quartile (55 stores, scores 37+)Bottom quartile (53 stores, scores 30 or less)

Most of the stores scoring 37 or more (47 of 55) are on the point-of-sale platform, which supplies much of this by default. The stores at the bottom are almost all on other platforms or custom-built sites (only 2 of 53 are on that platform). These features are part of what the assessment checks, so read this as "what the stores above the floor have in place", not as proof that one feature causes a score.

One group sits at the very bottom: 7 stores on Square Online scored between 7 and 9. Their storefronts build the whole page in the visitor's browser with JavaScript, so a reader that doesn't run JavaScript, as most AI crawlers don't, sees an almost empty page.

Finding 6: where the points go missing

Running stores trail ski shops in every category except product discoverability, and by the most in agent readiness (35 vs 58), the category that covers whether an AI agent can actually find prices, policies, and a way to buy. Prices that only appear after the page runs JavaScript are a large part of that.

The gap between what a site says and how it sells is wider too. The median store scores 53 on AI Search Visibility (what any assistant can read) but 8 on Your Own AI Agent (an AI that knows the store's products and can sell in conversation). Only 15% of running stores run a chat widget of any kind.

That second gap is the one our Shopify app is built to close, so read this paragraph with that in mind. The app gives a store its own selling agent, a digital twin of the shop: it knows every product, size and stock level, learns the store's own knowledge (fit advice, return rules for worn shoes, race-day hours), answers shoppers on every page of the store with product cards, and never recommends what's sold out. AI shopping agents can consult it on the store's Ask our expert page and through site tools. It's in early pilots now, and it only works for stores on Shopify, about a quarter of the stores in this benchmark.

Running vs ski, side by side

Same assessment, same four assistants, same scoring rules. A store on the same platform scores about the same in either category. Running stores score lower overall because so many of them sell through a storefront that caps what AI systems can read.

The national chains, for reference

We scanned four national chains alongside the independents. Fleet Feet scored 34 and Road Runner Sports 49. JackRabbit now redirects to Fleet Feet, so it has no separate score, and Running Warehouse couldn't be reached by our scanner. Fleet Feet's score sits right among the independents: being big doesn't make a store more readable to AI systems.

The stores we couldn't score

Why 21 listed stores weren't scored

Of the 304 independent stores on our final list, 282 were scored and ranked. We don't guess at the rest, so they're counted separately and never given a number:

• 12 had no online store: brochure sites or stores that sell in store only.
• 9 blocked our scanner on two attempts about a day or more apart. Our scanner runs from a cloud data center, and many sites refuse that traffic by default. It says nothing about whether a given AI assistant can reach the site.
• 1 never responded to our requests.

The top-scoring stores

Fourteen stores scored 70 or higher, the Mostly Ready band. Scores vary by a couple of points between runs, so we list them by score and don't rank within it.

None of them reached the top band. The next steps up are the same for every store: structured product data with prices on the page, organization data that tells a machine who and where you are, and an assistant that can answer product questions in conversation.

What any store can do this week

These line up with the biggest differences in the data. None of them requires sqrly.

1. If you sell through a shared storefront (a point-of-sale platform's online store, Square Online, or similar), ask your provider three things: does the storefront publish structured product data, are prices in the page itself rather than loaded by JavaScript, and does it offer an llms.txt? Their answer decides most of your score.
2. Publish structured product data on every product page: name, price, availability, brand, and a product identifier (GTIN/UPC) where you have one.
3. Add organization data to your homepage: name, address, phone number, and links to your social profiles.
4. Check your robots.txt for rules that block AI crawlers. Off Shopify, a quarter of the lowest-scoring stores (26%) block at least one.
5. Make sure you have an XML sitemap, and that robots.txt points to it.
6. Answer the questions runners ask (returns on worn shoes, gait analysis, group runs, race-day hours) in plain text on your site.

Is your store in the benchmark?

Every store we scored has its own report. If you run one of them and haven't seen yours, get in touch and we'll send your store's report and where it placed.

Not in the benchmark? Run a free sqrly Pulse snapshot on your own store: no account, no card. It takes a few minutes and shows what AI assistants can and can't read about your store, and what to fix first.

Become a pilot store. We're looking for a handful of stores where good advice decides the sale (on Shopify today, more platforms soon), and a running store, where fit is the whole conversation, is exactly that. Pilot stores get their own private install of the sqrly digital twin and work directly with the founder, and features they ask for can ship within days. Email florian@sqrly.ai with your store's URL.

Methodology

• Population. Independent US running specialty stores with an online store, compiled from industry directories and web search, then reviewed by hand. Big-box retailers, marketplaces, and brand-direct stores were excluded. Four national chains were scanned as a separate, unranked reference.
• Where we scanned. The site where each store sells. When a store's shop lives at a separate address (a "shop." subdomain or a hosted storefront), that's what we scored, so a brochure homepage doesn't stand in for the store.
• One build, one window. Every store was scored by the same assessment build between October 2 and 8, 2026, with the same four assistants as the ski benchmark. Changes made during the window affected only fact-check wording or whether a scan completed, never how a completed scan was scored.
• Client-rendered storefronts. A few Square Online sites first looked empty to our dead-site check because their pages are built in the browser. We rescanned them with that check switched off and scored them as a reader that doesn't run JavaScript sees them, the same way every other store is scored.
• What the score measures. The overall score blends AI Search Visibility (five categories, measured the same way for every store), Platform Shopping (live product-feed and agent checks, available only for Shopify and WooCommerce stores), and Your Own AI Agent. Assistant answers were collected live with web search on.
• Exclusions are counted, not scored, and listed above.
• Reproducible. Every number in this report is computed by a script from the stored scan data, and the charts are generated from the same data.
• Scanner update after the study (October 9, 2026). Shopify now publishes a default llms.txt file on every store automatically. Our scanner used to credit that file as if the store had written it; since October 9 it no longer does. The scores in this report are the benchmark snapshot from the build used October 2–8, and they are unchanged. On today's scanner, Shopify stores score roughly 4–5 points lower. Stores on other platforms aren't affected by this change.

How to read these numbers

• One scan per store. Scores vary by a couple of points between runs, so read bands and medians, not decimals.
• A snapshot. Assistant answers change over time and between sessions; these reflect what we observed in the scan window.
• Descriptive. These results show what the data looks like. They don't prove that any single change causes a higher score.
• About AI readability, not store quality. A low score says nothing about a store's shoes, fitting, or staff. It says how much of that an AI system can currently read.