Four ways a data deal gets priced

Owners tend to ask what their data is worth. The more useful question is how it would be paid for, because the structure moves the number as much as the dataset does.

The public deals give a sense of the range. Google pays Reddit a reported sixty million dollars a year. OpenAI reportedly agreed more than two hundred and fifty million dollars with News Corp across five years. Same industry, two different shapes: one an annual fee that renews, one a multi-year commitment paid against a fixed term.

Operational data has four structures available to it. Most owners only ever get offered one.

1. Subscription

A recurring fee for continuing access to data that keeps arriving. This is the structure that suits almost every operational dataset, because the value is in the fact that tomorrow's records will exist too.

It also aligns the incentives properly. The buyer pays for as long as the data stays useful, which gives you a reason to keep the feed clean and gives them a reason to keep paying. It is the only structure where a dataset can be worth more in year three than in year one.

2. One-off

A single payment for a historical extract. Appropriate when the value really is in the archive rather than the flow, which is rarer than people think, and common when a company is being wound up and there is no flow left.

Be careful here. A one-off looks attractive because the money arrives at once. It also ends your relationship with the asset. If the same buyer would have paid a subscription, you have sold a decade of revenue for one year of it.

3. Exclusivity premium

A higher price in exchange for being the only buyer, or the only buyer in a defined category.

Exclusivity is worth real money to a buyer whose edge depends on nobody else having what they have. It is also the structure most likely to be undersold, because owners agree to it without pricing what they are giving up. If your data would have supported four buyers, exclusivity has to beat all four combined, not just the best single bid.

4. Revenue share

You take a percentage of what the buyer earns from products built on your data.

This sounds fair and usually is not, for the same reason it is rare in licensing generally. You have no visibility into the buyer's accounting, no control over how they attribute revenue across inputs, and no ability to audit at a sensible cost. Treat any revenue share as a headline number to be discounted heavily.

Why the structure matters more than the headline

Two offers of the same size are not the same offer. An annual subscription at a modest figure, renewing, from a buyer who becomes dependent on the feed, is worth more over five years than a large one-off that closes the door behind it.

This is also why the number of interested buyers matters before you agree anything. A single interested party will propose the structure that suits them. Several interested parties turn structure into something you choose rather than something you accept.