Owner, operator, tenant: three different organizations
Ask who owns a stadium and you will usually get one name back. Ask again and you may get a different one, equally confidently. Both answers are often right, because the question is underspecified.
Take a typical large North American arena. A city or county holds the title to the land and the building. A management group holds a long-term operating agreement and runs everything from event booking to concessions. A club plays there under a lease and sells the naming rights, or does not, because the authority kept them. Three organizations, three different relationships, three different end dates.
What this dataset does instead
Ownership is stored as a list of dated roles rather than a field on the venue:
- owner, operator and tenant are separate role types
- each role row has a start date, an end date and a current flag
- a venue can have several current owners and several current operators at once
- every role row carries the URL it was derived from and a confidence level
That means a joint public and private ownership is representable without picking a winner, and a venue that changed operators in 2019 keeps both the old and the new row.
Why the table still shows one owner column
The table on this site has an Owner column and an Ownership column, and both are derived
conveniences, not the underlying truth. The owner column joins the names of current owners with a
comma. The ownership column reduces the set of current owners to one of four words:
publicif every current owner is a public authorityteamif the owner is the club itselfprivateif the owner is a companymixedif more than one kind of owner is current
Those derived values exist so the table can be filtered and sorted quickly from a compact index. Whenever they matter, follow the link to the venue page, where the full role rows with dates and sources are laid out.
The failure mode this avoids
The common shortcut, one owner string per venue, has a specific and predictable failure: it
records whichever organization the most recent article happened to mention. In practice that is
usually the operator, because the operator is the one issuing press releases. A dataset built that
way will tell you that a large management company owns dozens of publicly owned arenas.
Keeping the roles apart costs one extra table and a slightly more complex export. It is the difference between a dataset you can use for an ownership question and one you cannot.