Anglera · State of Product Experience 2026

The Supply Side

The industry spent 2026 asking 8,736 people what they need from a product page. Nobody checked whether the catalogs can deliver it. We measured 73 of them. They can’t.

11
Median attributes per page
Across 37 measured distributor catalogs
36/73
Refuse a plain web request
The only kind an AI shopping agent makes
8/73
Publish product structured data
At all, in any form

Get the designed PDF

The full five-page report. Work email — we send it and follow up once.

Every finding about what shoppers want is also a bill for work nobody did

The demand-side research is good and it is half a study. It asks shoppers what they want and stops before anyone checks whether the shelf delivers it. Left column, what they said. Right column, what we measured.

83%of online shoppers are likely to abandon a site when they can’t find the information they want.
Median attributes on a distributor product pageMeasured · n=37
11
Ten of 37 catalogs ship five or fewer. Ferguson’s carry none we could read.
52%rank detailed descriptions of features and capabilities in their top three purchase factors.
Descriptive content, scored out of 10Measured · n=37
4.5
Imagery scores 4.8 of 10. Identifiers, 3.8 out of 12.
80%think less of a brand whose products are poorly represented online — a fourth straight annual rise.
Widest quality spread inside one catalogMeasured · n=37
45 pts
Same company, same shelf. The median catalog varies 11 points page to page.
“AI”is the first audience that enforces every standard that has been neglected, automatically and at scale.
Storefronts publishing product structured dataMeasured · n=73
8/73
And 36 of 73 refuse a plain web request. 9 publish a sitemap inviting the crawler they then block.
53%rank ratings and reviews the most-selected purchase factor, beating discounts outright.
Reviews have no column in our indexNot measured
In distribution the shortlist is settled on specification, not sentiment — which returns you to row one.

Left column: Syndigo, The State of Product Experience 2026 — 8,736 adults across six markets, fielded by YouGov, June 2026. Right column: Anglera, Top Distributors 2026 — 37 catalogs sampled page by page and 73 storefronts retrieval-tested, August 2026. Full method.

Ways a catalog says no to a machine that came to buy

None of these is a slow site or a bad site. Each is a specific, identifiable decision — and each is invisible to the people who made it, because a browser renders around it.

Linde (Americas)

shadow dom listings

Coveo shadow-DOM listings; product links are not in the served document at any depth.

Henry Schein

bot wall on public pages

Akamai bot wall in front of pages that are fully public. Login gates pricing, not the pages — so the wall blocks machines from content humans read freely.

Border States

empty top categories

Top-level category pages list only subcategories and contain no products; products live one level down.

Consolidated Electrical Distributors (CED)

no national storefront

No national catalog exists. ced.com refuses every client, myced.com is SharePoint behind SSO, and the storefront is replicated per profit center across 700+ *.portalced.com banners with no root domain.

The Hillman Group

links are not links

Every product tile renders as href="javascript:void(0)" and navigates in JS. A crawler following links finds 0 products; the site's own sitemaps carry ~64,590 PDP URLs.

McNaughton-McKay Group

client rendered shell

Insite SPA. Category grids render client-side, so served HTML is an empty "loading content" shell. Its public /api/v1/products serves the same data.

9 of the 26 storefronts that publish an XML sitemap refuse a plain client. A sitemap exists to tell an automated visitor where the products are. Publishing one and then blocking the visitor is not a security posture — it is a contradiction.

Nobody in this industry is short of software. They are short of the work.

Twenty years of tools — MDM, PIM, syndicators, agencies — gave this industry places to store product data and pipes to move it. The work always landed back on the same small team.

That is the whole finding. A PIM holding eleven attributes publishes eleven attributes. A syndication feed carrying a page with no GTIN carries a page with no GTIN, faithfully, to every channel. None of the failures above is a storage failure. Every one is a job that was never staffed.

Which is why the demand-side research keeps landing without changing anything. Telling a distributor that 83% of shoppers leave when information is missing is not new information — it is a bill. Somebody still has to open 60,000 spec sheets and fill the fields.

Your PIM stores the data. Anglera does the work.

What doing the work actually produces

100,000+
SKUs enriched in one enterprise pilot — 2.6× the contracted scope
~95%
found and fully enriched across two master catalogs
15–60+
structured attributes filled per product, by category
800,000+
attribute-level source citations delivered
81%
of values extracted directly from a cited source, not generated
Overnight
10,000 SKUs enriched in a single run

One anonymized enterprise-distribution pilot. Elsewhere: 60,000+ products cleaned and enriched in a first month at a trade distributor, 12,000 of them live on site inside two weeks; 23,000+ products rebuilt at Cresco Resco by one person working part-time.

Find out what your own shelf ships.

Every company in the index has a public scorecard — four pillars, thirteen signals, page-level evidence. Yours takes one URL.