The Resolution Layer · Case Study in BriefEKOM
A national furniture retailer.
A no-schema read of a marketplace-bound catalog — one of a set of anonymized EKOM analyses. What surfaced before the listings could clear, and where it leads.
Lead finding · legal but untrue
Filled fields that are plainly wrong — and a gate that waves them through.
A dining table lists Aluminum three times, summing to 300%; other items carry material percentages over 100%. Origins contradict the shipping port; Prop 65 warnings are still unfilled placeholder text; the identity product_sku is blank on nearly every item. Each is populated and validly formatted — so it passes the marketplace's ingestion check, which confirms fields are present and formatted, not that their values are true.
Also surfaced
Fails marketplace validation
Required fields blank — LegMaterial, ProductName, Height, carton dimensions — so the listing is flagged INVALID and can't go live at the ingestion gate.
Fulfillment & freight blockers
Missing height and carton dimensions break freight quoting; a shippable item lists Volume = 0; bed frames typed as kits with no components — unfulfillable as listed.
Pricing & routing integrity
Active dropship products with $0 cost, a blank dropship cost, and a special-order lead time of 1,000 days sitting as an unremoved placeholder.
Excel-corrupted data
First-inventory dates of 12/31/1899 across many items — a spreadsheet epoch null that reads as complete and is effectively empty.
How EKOM read it. The catalog is small enough to read whole but dense with per-item defects, so the engine ran a granularity ladder — one broad pass plus three focused passes — with no schema and no manual setup. Each focused pass out-found the broad one (18 / 23 / 16 vs 15): a small, attentive pass catches the impossible percentages and epoch dates a single sweep averages away.
From diagnosis to channel-ready. This pass read and diagnosed. The same understanding powers what follows — correcting the impossible math, origin contradictions, and placeholders, populating the identity key, filling the required fields the marketplace rejects on, and holding the line at intake so new vendor loads come in clean and marketplace-ready.