How we decide

1. The brand database

We keep a database of private-equity firms and the consumer-facing brands they own, built from two sources:

• A curated seed list of roughly 90 well-documented brands — restaurant groups, dental and vet roll-ups, home-services franchisors, car washes, gyms, retail — each with a source link and an as-of date. Because these are hand-verified or drawn from reported deal coverage, they carry a higher ownership-sourcing weight (verified 1.0, or reported 0.85 — see the confidence formula below).
• Scraped PE portfolio pages — firms’ own public “our companies” pages, scraped directly. These entries carry a lower ownership-sourcing weight (0.7), since a portfolio-page listing is thinner evidence than a sourced, verified claim.

2. Nearby businesses

For whatever area you’re viewing on the map, we fetch nearby businesses live from OpenStreetMap’s Overpass API — the same free, public map data anyone can query. Results are cached for 24 hours so repeat searches of the same area are instant. Coverage is only as good as OpenStreetMap’s completeness for that area.

3. Name matching

Each nearby business is compared against every brand name (and known aliases) in the database. Names are normalized first, then matched by exact match, by one name containing the other, or — for close-but-not-exact names — by fuzzy (Jaro-Winkler) string similarity. Only high-confidence fuzzy matches count.

OpenStreetMap sometimes tags a location with an explicit brand field, which we prefer over the visible storefront name when both match — a mapper’s brand tag is usually more deliberate than a free-form name. But brand tags can go stale after a location changes hands (a new tenant moves in and the map isn’t updated yet), so if the brand tag doesn’t agree with the visible name at all, we treat it as suspect: the match’s confidence is cut by 40% before the confidence cutoff is applied.

4. The confidence score

Every match gets a confidence score from 0–100, built from three factors multiplied together:

• Name-match quality — how sure we are the storefront name refers to the brand (an exact match scores highest; a fuzzy match scores lower).
• Ownership sourcing — how solid the evidence is that the brand is PE-owned: verified ×1.0, reported ×0.85, scraped ×0.7.
• Stake type — how directly the firm controls the brand: majority owner ×1.0, franchisor ×0.95 (each location is typically an independent franchisee), minority stake ×0.75.

Matches scoring below 55 are dropped rather than shown — below that line, a name match is more likely coincidence than an actual ownership link.

Showing our work

Every ownership claim lives in an entity graph, where each connection carries its own source type, source URL, as-of date, and confidence. Click any marker on the map and open “The connection” to see the full evidence chain: how the storefront matched the brand, who owns the brand and per what source, the firm’s SEC investment-adviser registration, and the funds it raises capital through, each linked to its actual SEC filing.

Ownership changes constantly — treat every match as a starting point for your own research, not a fact.