How to sell data your company holds

Very little company data is sold outright, the way you would sell a laptop. Most deals are licences: the owner keeps the data and grants a buyer the right to use a copy for stated purposes, for a period, sometimes exclusively. That distinction matters from the first conversation, because it decides what you are promising, what you can still do with the data afterwards, and what the buyer will expect to inspect.

This guide is for an organisation that holds data it built for its own work, such as a software company with years of code, a logistics firm with shipment records, or a publisher with an archive. It is not about individuals selling their personal browsing data to apps, which is a different market with very small payments.

1. 1

   Make an inventory of what you hold

   Write down each dataset: what it describes, how it was collected, the date range, the size, the format and who maintains it. Owners often undercount here. Old product databases, support tickets, internal documents and abandoned codebases can matter more to a buyer than the dataset you think of first.

2. 2

   Trace the rights for each dataset

   For every item, find the documents that say who owns it and what may be done with it: your terms of service and privacy policy at the time of collection, contracts with customers or partners, contributor agreements, and any third-party licences. If data came from a customer or a vendor, you may not be free to license it, however long you have stored it.

3. 3

   Deal with personal information

   Decide whether personal data has to be removed, de-identified or kept with a lawful basis for sharing. Removing it is usually the simplest route and widens the pool of buyers. If it must stay, take advice from a privacy lawyer before any sample leaves the building.

4. 4

   Package and document it

   Write a data dictionary (every field, its meaning and its units), explain how records were collected and what gaps exist, and cut a small representative sample. A buyer who can understand a dataset in an afternoon moves faster and argues less about price.

5. 5

   Set a price range, not a number

   Think about what it would cost the buyer to collect the same data, how scarce it is, how fresh it is, and whether they get it exclusively. Write down a low and a high figure and what would move a deal between them. A first estimate is a starting point for conversation, not an offer.

6. 6

   Choose how you will reach buyers

   You can list on a data marketplace, work with a broker who brings buyers and takes a fee, or approach buyers directly. Each suits a different kind of dataset; the guide on where to sell data compares them.

7. 7

   Sign confidentiality before sharing detail

   Before a buyer sees a sample or a schema in depth, sign a non-disclosure agreement. Once there is serious interest, agree the main terms in a short letter of intent: scope, price, exclusivity, delivery and timing.

8. 8

   License, deliver and get paid

   The data licence agreement sets permitted uses, restrictions on resale, warranties about rights, liability limits, delivery format and payment terms. Deliver through a controlled channel such as a data room, keep a record of what was sent, and confirm the buyer accepts it against the sample.

Fill this in for each dataset before you speak to a buyer. Any row you can't answer is a question a buyer's lawyer will ask later, at a worse moment.

• Pitching before checking rights. The most expensive way to learn your customer contract forbids resale is from the buyer's lawyer.
• Sending raw data as a "sample" before an NDA. Once it has left, you can't take it back.
• Quoting a single price. A range with reasons survives negotiation; a number invites a counter-offer that anchors low.
• Promising exclusivity casually. An exclusive licence can block every other sale of that data, so price and scope it deliberately.
• Forgetting updates. Some buyers only want a one-off snapshot; others need a feed. Say which you are offering.

DataBroker is a course and a set of tools for brokers, the people who sit between data owners and buyers. If you own data and would rather not run the sale yourself, a broker can do the sourcing of buyers, the paperwork and the negotiation for a commission agreed in writing. If you are the broker, the app's estimations give a first value range for codebases, task data and raw data, and its contract templates cover the mutual NDA, letter of intent, data licence and broker agreement. Neither replaces your own lawyer.