← Legibility Index 2026

The Legibility Gap

What a retailer publishes and what a machine can actually recover from the same page are two different things. We measure both, and the distance between them is the most useful number on the scorecard.

Two ways to fail, one score

Read a product page as a plain crawler and you see the structured data the retailer chose to publish. Run the same page through an extractor and you see what a machine can pull back out of it. Those disagree far more often than anyone expects, and they disagree in two opposite directions.

Published well, recovered poorly

The markup carries a valid GTIN, an MPN, a return policy. The extractor gets the name and the price and drops the identifiers — the exact fields an agent needs to match the product to the same product at a competitor. The retailer did the work and gets none of the benefit.

64of 111 scored retailers

The data exists, the door is shut

The page is rich, but a standards-compliant client gets a 403. Bot protection that stops a scraper also stops the crawlers behind the assistants the retailer wants to be quoted by. We report this as unreadable, never as empty — we do not claim to know what we were not shown.

15of 163 retailers

Both tracks agree

The retailers whose markup survives extraction intact. Whatever their score, these are the catalogs an agent can actually use as published.

26of 111 scored retailers

Where identifiers are being lost

These retailers publish identifiers that did not survive extraction on at least one sampled page. It is usually the cheapest fix on the whole scorecard, and the highest-leverage one.

RetailerScoreLost in extractionPages
Scheels74gtin, mpn7 of 7
Ace Hardware73mpn8 of 8
Academy Sports + Outdoors72rating5 of 8
Tapestry (Coach)71mpn, gtin8 of 8
Blain's Farm & Fleet70gtin, mpn, rating8 of 8
Walgreens69gtin, rating8 of 8
Costco Wholesale67rating5 of 5
The Home Depot67gtin, rating8 of 8
Sam's Club67rating, gtin8 of 8
PetSmart66gtin, rating7 of 7
REI Co-op66gtin7 of 8
Sportsman's Warehouse66mpn, rating7 of 8
Dillard's65rating5 of 8
IKEA (North America)65rating6 of 8
Carter's65mpn, rating7 of 8
Guitar Center65gtin, brand, rating6 of 7
Camping World64mpn, rating6 of 7
Signet Jewelers64mpn, brand8 of 8
AutoZone63mpn8 of 8
Floor & Decor63brand2 of 8
Crutchfield63mpn, rating8 of 8
Nike62mpn8 of 8
B&H Photo Video61rating2 of 3
Ahold Delhaize USA60rating, brand8 of 8
Dell (Direct)60brand, rating, mpn7 of 8
Michaels60gtin2 of 7
Zappos60rating6 of 8
World Market60mpn, rating8 of 8
Discount Tire59rating2 of 7
Harbor Freight Tools59rating7 of 7
Staples59rating4 of 7
Northern Tool + Equipment59mpn, rating, brand7 of 8
Etsy59rating1 of 1
Urban Outfitters (URBN)58mpn, rating8 of 8
Anthropologie58mpn, rating8 of 8
Hy-Vee57gtin7 of 7
Bath & Body Works57rating7 of 7
True Value57gtin, mpn5 of 5
CarParts.com57mpn2 of 2
Macy's56rating8 of 8
Foot Locker56sku8 of 8
Lowe's55rating1 of 3
Ulta Beauty55rating3 of 8
Hobby Lobby55rating5 of 8
QVC Group55brand1 of 8
Advance Auto Parts54rating1 of 1
Abercrombie & Fitch54rating6 of 8
Barnes & Noble54rating1 of 8
Dollar General53rating4 of 8
Tractor Supply Co.53rating3 of 4
AT&T (Retail)52rating1 of 8
Defense Commissary Agency52gtin6 of 6
Dollar Tree51brand4 of 8
The Container Store51mpn, rating3 of 4
Gap50brand, rating6 of 6
Victoria's Secret49rating2 of 5
Micro Center48mpn, rating1 of 1
CVS Health47rating2 of 6
Pep Boys46mpn, brand7 of 7
Total Wine & More45rating5 of 8
Belk44rating3 of 3
Albertsons43gtin1 of 2
Amway40brand7 of 8
Casey's General Stores39sku, brand4 of 4

Why we can measure this

Because running an extractor across other people's catalogs at scale is what Anglera does. The second track is not an add-on to the benchmark — it is the same machinery we point at a customer's own supplier data, turned around and pointed at the open web. That is why this comparison exists here and not in the other retail rankings: it takes an extractor to see it.

Across the scored list, 0% is the median share of sampled pages carrying a GTIN that actually validates its check digit — a field that says gtin but does not validate is worse than none, because it silently breaks the cross-retailer matching it was supposed to enable.