The Resolution Layer · Case Study in BriefEKOM
Stainless steel production.
A product-data read of a stainless steel producer's public website — product pages, product sheets and grade data sheets read side by side — where the same steel is described three times and the three do not always agree. One of a set of anonymized EKOM analyses — what surfaced, and where it leads.
Lead finding · a grade designation the producer uses nowhere else
A lean duplex grade, published on one product line under a designation the producer uses nowhere else.
The page and its product sheet both print the right producer code and the right UNS number — and beside them a type designation that is not a standard duplex designation. Every other page that sells the grade, and the grade's own data sheet, get it right. On the same product sheet, the grade table has slipped: the type row no longer lines up with the UNS row, so four rows name the wrong steel and a fifth has lost a digit. A buyer searching by grade name misses the product, and a buyer reading the sheet is told that 316L's UNS number belongs to two other steels.
Also surfaced
Data sheets that disagree with the producer's other documents
Four data sheets — two duplex, two ferritic — print maximum carbon at ten times the low-carbon limit, while another of the producer's sheets prints the right one for both duplex grades. A European number that does not match the steel name beside it; one grade with two chromium ranges on two sheets; a designation with a dropped digit, and two with garbled letters.
A page and its own PDF
Seven times, a product page and the sheet it links publish different size ranges or grade lists for the same product — a bar range that starts lower on the page than in the sheet; a duplex bar in one grade or two.
Grades sold, not documented
Ten grades offered on product sheets have no grade page and no data sheet, and the flat-product pages carry no full grade list — the grades sit in full only in the PDFs.
Context: the surcharge is clean
Every row of the six alloy-surcharge tables read, across two months, adds up to its stated total to four decimal places. The document republished every month is the one that agrees with itself.
How EKOM read it. The read took the public site as published — no file handoff, no credentials — and lined up each product's grades so the web page, the product sheet and the grade's data sheet sit side by side, keyed on the producer's own grade code and the UNS number. Every PDF finding was checked on the PDF itself, and the headline values were re-read on the live site before this study was published.
From diagnosis to one grade record. The fifteen findings point to four likely mechanisms — three documents kept separately, tables retyped rather than generated, one sheet copied to make others, and copies not drawn from one grade record. Fixing the lines fixes these sheets; generating every page and sheet from one record, and checking each against it before it is published, fixes the next reissue before it goes out.