The Resolution Layer · Case Study in BriefEKOM
Mattresses & bedding.
A cold read of the mattress catalog of a large home-furnishings retailer — where the shopper buys by one word, Queen or King, and that word is carried by two numbers that have to be the right way round. One of a set of anonymized EKOM analyses — what surfaced, and where it leads.
Lead finding · dimensions that follow ownership
Width and depth are reversed on every brand the retailer owns — and almost nowhere else.
374 mattress listings publish the head-to-foot length as the width: an own-brand queen reads width 80, depth 60, where a national-brand queen reads 60 by 80. The reversal is near-total on each of the retailer's own brands and the exception on the national brands it resells — none at all on four of them. Hundreds of people typing dimensions do not agree to be wrong along a line of corporate ownership; a mapping does. That makes the largest finding in the analysis the cheapest to fix: most likely one correction per affected feed, toward a convention the same catalog already follows on the brands it resells.
Also surfaced
A series in which every listing carries another bed's measurements
Across one national-brand series the size in the name and the size in the numbers never agree — a California King shown with a Queen's footprint, a King with a California King's, a Twin XL with a Full's. Every value is a valid size, just not the one named, so the correct pair is already known.
Groups with the specification missing
Fifteen premium split California Kings publish a width and nothing else, though each one's height is written into its own name. Five listings carry no price and no availability, so no cart and no shopping feed can take them. One size in a six-size family reads five inches thicker than the rest.
One product at two addresses, many products under one name
Forty-one mattresses are published at two URLs, identical on every field and both in the sitemap. Forty-eight names are shared by more than one product — twelve distinct products under a single name — and one series leaves the size out of the name on sixteen listings.
Nothing a machine can filter on
There is no field for size, firmness or construction — the three filters mattress shoppers use. A size filter has to lean on dimensions reversed across the owned brands, or on names that sometimes omit the size. Underneath, three dead pages are served as live, and every product in the sitemap carries the same priority.
How EKOM read it. The engine took the live public catalog and its sitemap cold — no file, no credentials, no schema — profiled the catalog's own field semantics, then held every record against its siblings, its series and its size. That context is what turns a plausible number into a finding. Here it was the national brands: they carry width and depth one consistent way, and that convention is the only reason a reversal on the other brands is visible at all. Nothing was told to the engine about which brands the retailer owns; the defect drew that line on its own.
From diagnosis to channel-ready. The 16 findings collapse into six upstream mechanisms — an orientation set wrong per feed, a size's dimensions attached to the wrong variant, variant groups dropping fields in transit, products published twice, a specification that lives only in the name, and defaults and internal values standing in for real ones. Fixing the findings fixes the findings; fixing the mechanisms fixes every mattress that has already passed through them and every one that will, including the next brand the retailer adds. Then the same intelligence holds the line at intake, so the next feed that arrives the wrong way round never reaches a live page.