Buy datasets: where to get data and what to check
Buy datasets from the source that fits the job: free catalogues, marketplaces, providers, or straight from the company that holds the data.
Updated
All buying guides
- Updated Nov 9, 2026Buying consumer data
Buy consumer data only after five checks: the seller's registration, the data's source, sensitive categories, your purpose and who else gets it.
Where to buy datasets, compared
Compiled by DataBroker from the sources below (checked 2026-11-09).
Public catalogues (Data.gov and state portals)
- What you get
- Government datasets, usually free downloads
- The licence
- Open terms set by each publisher
- Best for
- Baselines, demographics, public records
- Check first
- Whether the file is current and what each column means
Dataset search engines (Google Dataset Search)
- What you get
- Links to datasets hosted in thousands of repositories
- The licence
- Whatever the host sets; the search engine doesn't host the data
- Best for
- Finding out what exists before paying
- Check first
- The licence on the host's page, not the search result
Data marketplaces (for example AWS Data Exchange)
- What you get
- Subscriptions to listed products, from 1 to 36 months on AWS, by file, API or data share
- The licence
- The provider's data subscription agreement, accepted before you subscribe
- Best for
- Standard commercial data you can trial and switch
- Check first
- The agreement's permitted uses, renewal terms and the provider's refund policy
Data providers and brokers
- What you get
- Compiled files and feeds on companies, people or places
- The licence
- A negotiated or standard licence
- Best for
- Marketing, risk and enrichment data at scale
- Check first
- Whether a seller of personal data is registered where it must be
Direct from the company that holds it
- What you get
- Data no one else sells: logs, archives, code, transactions
- The licence
- A licence you negotiate, often through a broker
- Best for
- AI training and niche analytics
- Check first
- That the seller owns it and may license it, and what was removed
Sources
5 sources · checked Nov 9, 2026
More about buy datasets
Before you buy datasets
Buying a dataset almost always means buying a licence: the right to use a copy for stated purposes, for a period, under the seller's terms.
The same records can be free for research and expensive for commercial use, so decide what you will do with the data before you compare prices.
Then check whether it already exists for free: Data.gov alone listed 532,497 datasets on 9 November 2026.
What the licence should say
- Permitted uses, named plainly: analytics, model training, resale, display to customers. Anything not listed is a dispute waiting to happen.
- Term and what happens at the end: delete the data, keep derived work such as a trained model, or keep everything.
- Exclusivity, if you are paying for it, and for which field or region.
- Provenance: where the records came from and the seller's promise that it may license them.
- Personal data: what was removed or de-identified, and who is responsible if something slipped through.
- Updates and delivery: how often, in what format, and what you can do if a delivery is late or wrong.
On a marketplace
On a data marketplace most of this sits in the provider's agreement; on AWS Data Exchange, for example, you must accept the provider's data subscription agreement before you can subscribe, and refunds follow the provider's own policy.
Checking the seller
For data about companies or machines, the main question is ownership: did the seller create the data, and has it licensed it exclusively to anyone else? For data about people, add the law.
When the data is about people
A company selling personal information about people it has no direct relationship with usually has to register as a data broker in California, Vermont, Texas and Oregon; you can check whether it is a registered data broker in California's 2026 list.
If the data will be used to decide on credit, insurance or employment, it is a consumer report and the Fair Credit Reporting Act applies to both of you.
Buy datasets FAQ
Where can I buy datasets?
From public catalogues such as Data.gov (often free), data marketplaces such as AWS Data Exchange, data providers and brokers, or directly from the company that holds the data. Search Google Dataset Search first to see what already exists.
Can you buy data legally?
Yes. Most data is sold under a licence that sets what you may do with it. Personal data adds rules: state data broker registration for sellers, the Fair Credit Reporting Act for eligibility decisions, and privacy laws on sale and sharing.
Do I own a dataset I buy?
Usually not. You buy a licence to use a copy for stated purposes and a stated term. Outright transfers of ownership happen, for example when a company sells its assets, but they are the exception.
Are there free datasets I should check first?
Yes. Government catalogues publish hundreds of thousands of datasets; Data.gov listed 532,497 on 9 November 2026. Free data rarely gives an edge on its own, but it is the baseline a paid dataset has to beat.
Related
- Data marketplace
A data marketplace lists datasets from many sellers in one catalogue, so buyers can search, sample and subscribe without separate negotiations.
- Best business data providers
Best business data providers grouped by data type, with who buys each kind, how it is priced and which rules apply.
- Data valuation
Data valuation explained for people selling or brokering a dataset: three standard methods, what each misses, and a worksheet to score one.
Take the DataBroker course
The DataBroker course and tools: find companies with data to sell, price it, license it and close the deal.