Data Marketplace Guide

Home / How to buy a dataset

How to buy a dataset

A checklist for buyers: from a clear brief to a license you can defend.

Last checked 11 November 2026 · General information, not legal advice

On this page
  1. 1. Write the brief
  2. 2. Shortlist sellers
  3. 3. Test a sample
  4. 4. Check provenance and rights
  5. 5. Read the license
  6. Questions to ask the seller
  7. Questions

1. Write the brief

State the question the data must answer, the fields you need, the history required, the geography and the update frequency. A vague brief produces vague samples.

2. Shortlist sellers

Search the marketplace that matches your platform first, then a directory such as Datarade for sellers outside it. The comparison shows how they differ. Shortlist three to five.

3. Test a sample

Run the sample through your real pipeline. Check coverage, duplicates, missing values and how it joins to your other data. A sample that looks good in a spreadsheet can still fail in production.

4. Check provenance and rights

Where did the records come from, and does the seller have the right to license them? This matters most for anything derived from people, such as messages, tickets and transactions. A weak answer here is a reason to walk away.

5. Read the license

  • Permitted use, including model training and derived outputs
  • Term, renewal and what happens when it ends
  • Exclusive or non-exclusive
  • Who owns derived data and models
  • Indemnity if the seller had no right to license

Questions to ask the seller

  1. How was the data collected, and under what permissions?
  2. How was personal data removed or protected?
  3. How often is it refreshed, and for how long will it be supported?
  4. Can I see the full schema and a data dictionary?
  5. What is the delivery method and format?
  6. Who else holds this data, and is any of it exclusive to me?

Questions

Can I use a purchased dataset to train an AI model?

Only if the license says so. Many licenses limit use to analysis or internal purposes. Ask for the permitted use in writing and confirm it covers training and any resale of outputs.

How do I know the data is legitimate?

Ask where it came from, whether the seller has the right to license it, and how personal data was handled. Ask for written confirmation, not a verbal answer.