How scores are built

Two scores, one scale

Each run produces an existing-card score (0–100) and a new-line score (0–100). The first looks only at the published book as submitted. The second looks at how a candidate factory or principal sits against that book.

Bands used by this prototype: 85–100 exceptional coherence on the published facts; 70–84 strong; 55–69 moderate; 40–54 limited; 0–39 high friction or conflict on those facts. Same bands appear on the public index.

Inputs

A current published line card (URL and/or pasted principal list). A candidate factory or principal (URL and/or short product description). Optional territory notes, customer types, and number of outside people.

When URLs are given, this prototype retrieves the public page and extracts visible text. It does not log in, scrape behind paywalls, or invent brands that are not in the intake or the page.

Factors

• Shared-call potential (do the published customers and categories look like the same visit?)
• Technical overlap (does the candidate sit next to named categories, or in a different stack?)
• Conflict / exclusivity risk from public language (same slot, competing catalogs, “exclusive” claims on the page)
• Card load (how crowded the published list appears)

What a score is not

Not a legal review. Not a sales forecast. Not permission to sign. Not compensation advice. Not an exclusivity opinion. Public pages can be stale or incomplete. The model can be wrong.

Disabled categories

This prototype is not built for process fluids, specialty chemicals, metalworking fluids, lubricants, or food ingredients. If those categories appear in the intake, no report is produced.

Focus of examples and the public index: metal components, electro-mechanical systems, thermal products, aerospace/defense hardware, medical devices, industrial OEM, electronics hardware.