Counting how many states an operator truly spans is entity resolution, not arithmetic, because registries see legal names while the market sees brands.
"Operates in a dozen states" is the kind of claim that sounds like a number and behaves like an argument. Depending on who is counting and what they are counting, the same company can span three states or ten — and every version can cite something real. Multi-state footprints resist clean counting for reasons built into how licensing works, and the mechanics are worth walking through, because they explain why registry-based counts and press-release counts so rarely match.
The first problem is the unit of count. A footprint could mean licenses held, locations operating, states entered, brands present, or entities controlled — five different tallies that headline writers compress into one. A company with many licenses in one state and a single license in each of two others is simultaneously a large operator and a narrow one. Until the unit is stated, the number is unanchored.
The second problem is that state law actively fragments corporate identity. Ownership rules, residency requirements, and disclosure regimes differ enough that multi-state companies routinely hold each state's licenses through separate subsidiaries with separate legal names. The registry in one state lists one entity; the registry in the next lists another; nothing on the face of either record connects them. Exact-name matching across registries — the most auditable method available — will find only the operators that kept their naming consistent, which makes any exact-match count a floor. Loosen the matching to catch near-names and the opposite risk appears: unrelated businesses that happen to share generic names get glued together into phantom operators. Every matching method trades one error for the other, and the honest ones say which trade they made.
Third, presence is not always ownership. Brands travel through licensing deals and management services agreements, letting a name appear on shelves in a state where the company behind it holds no license at all. Is that a footprint? By brand visibility, yes. By registry record, no. Both answers are defensible; a count that mixes them without saying so is not.
Fourth, the records that would settle ownership questions are unevenly public. Some states publish owner and officer information with each license; others publish only the entity name; a few publish rosters so minimal that even the entity name arrives abbreviated. Cross-state ownership analysis is only as strong as the most reserved state in the chain allows it to be.
Then there is timing. Footprints change through acquisitions, wind-downs, and license transfers, and each state's registry reflects those events on its own lag. A company mid-acquisition can legitimately appear in two states' rosters under two ownership descriptions at once. Any national count is therefore a mosaic of snapshots taken on different days — accurate tile by tile, and still capable of misleading in aggregate if the dates are not carried alongside the names.
Headlines inherit none of these constraints, which is the last reason the numbers diverge. Announcement counts tally intentions — states entered via deals announced, markets planned, footprints described in investor language — while registries tally records. The two can differ in both directions: a company may operate more broadly than exact-name matching reveals, and less broadly than its own communications imply. Neither source is lying; they are answering different questions on different clocks.
None of this makes footprint counting pointless. It makes methodology the story. A count worth using states its unit, names its matching method, admits the direction of its likely error, and dates its snapshots. A count arriving without those four disclosures is a vibe wearing digits.
The structural takeaway survives every methodological hedge: registries organize this industry state by state because the law does, and any national view is an act of assembly, not observation. Whoever does the assembling makes judgment calls, and the only trustworthy assemblers are the ones who show them. When a footprint number appears without its method, the useful response is not belief or doubt. It is a question: counted how?
Analysis: Platform Data Desk.