The Resolution Layer · Case Study in BriefEKOM
A sleepwear and loungewear brand.
Two full-catalog reads, one month apart — one of a set of anonymized EKOM analyses. What the first found, what the brand did with it, and what a blind second pass found underneath.
Lead outcome · pass two
They fixed it — and did it to every affected product, not a sample.
The brand implemented four of the six changes EKOM had fully specified, and left every item marked as needing their own judgment exactly where it was. On the one they took on most broadly, they resolved it in the opposite direction from the recommendation — normalizing the whole catalog to the convention that suited them, across every surviving product of both affected types. Nothing dropped out of a collection. That is a team that understood the findings rather than executing them.
What the second read found
A defect class that shrank without anyone fixing it
On every headline measure the color-capitalization problem looked half-solved. It wasn't touched — the affected items were legacy and sale stock that retired in the ordinary seasonal turnover. Of the products still published, not one was corrected. Scoring only survivors is what caught it.
Two colorways holding each other's color
Two colorways of one sweater have their color values swapped — titles, photography and swatches all correct, only the field the color selector reads is wrong, in both directions. A shopper picking the first is wired to the second.
Structured data that carries price and no product facts
The machine-readable block every product page publishes — what shopping feeds and AI assistants read — carries price, availability and size, and no material, pattern or category at all. For a brand whose product is print and fabric, that is the whole proposition going unread.
Spec content that exists and never reaches the feed
Composition, care and origin are published at better than 97% coverage — held in storefront metafields the product feed doesn't expose. The facts already exist, fully maintained, one mapping away from everything downstream.
Where EKOM corrected itself — twice. In pass one, the highest-severity pattern the engine produced was checked against live variant records, disproved and removed before anything was written down. In pass two, EKOM retracted its own characterization of the brand's spec content as absent; it was there all along, invisible to an instrument reading only the feed. A second pass is only worth commissioning if it is allowed to contradict the first.
From diagnosis to channel-ready. Two passes in, the shape of the work is clear: diagnosis is the cheap part. Four findings resolved in four weeks, four still open, and a new layer underneath — because a catalog that replaces half its range in a month re-applies every convention from scratch on every drop. What it needs isn't another audit. It's a standard that holds between them.