Data licensing guide

Data monetization for established companies: the models compared

Quick answer

Data monetization means turning data into measurable value. MIT CISR groups it into improving your work, wrapping data around products, and selling information, and found selling brought just 18% of returns. Licensing records to AI programmes is a newer form of selling: no product to build, but it depends on rights, and nothing is guaranteed.

51%
of data monetization returns come from improving work (MIT CISR)
31%
come from wrapping data around products
18%
come from selling information
No product
to build if you license a copy

Most people hear "data monetization" and think of selling data. The researchers who've studied it longest say that's the smallest part. We run a referral service for one form of it, licensing to AI, so we have a stake in this. Here's the wider picture, including where licensing isn't the right answer.

01The models, side by side

MIT's Center for Information Systems Research has studied this since at least 2014. Its 2023 book reports that on average 51% of data monetization returns come from improving, 31% from wrapping, and 18% from selling. Those figures describe companies' results overall and predate today's AI licensing programmes.

ModelWhat you doWho paysWhat it takes
ImprovingUse data to work better: cut waste, speed up decisionsNobody. You save moneyData skills inside the business
WrappingAdd data features to a product you sell, such as dashboards or benchmarksYour customers, through a better productProduct and data teams
Selling an information productSell reports, feeds or analytics, usually on subscriptionNew customersA data product and someone to sell it
Listing on a marketplacePublish a dataset for subscribersSubscribers, through the marketplaceClean, repeatable data and upkeep
Licensing a copy to an AI programmeGrant agreed use of existing recordsThe programmeYears of records you control, and a rights review
BarteringSwap data for a service or better termsPaid in kindA partner who values it

MIT's example of selling is a health care company that sells hospitals information on their spending patterns, so they can buy better. AWS Data Exchange is the marketplace kind: it invites providers to "reach AWS customers with your data files, tables, and APIs". Both need something a one-off licence doesn't: a repeatable product with customers to serve.

02Where AI licensing is different

Wrapping, selling and marketplaces all need your data to be a product. AI programmes want the opposite: the unpolished record of how work really got done, such as tickets, SOPs and project histories. So a company with no data team can take part.

Choosing a data monetization model: if you have a product and data team, consider wrapping or selling an ongoing data product. If not, and you hold years of your own records, licensing a copy is the lightest route. If you hold neither, start by improving your own operations with your data.YesNoYesNoProduct and datateam?Wrap a product, orsell a data productYears of records youcontrol?License a copy to aprogrammeStart by improvingyour own work
A starting point. Many companies do more than one.

What you trade for that simplicity:

  • No build, but no moat. A licence creates no product, and it won't improve your operations.
  • Irregular income. Programmes publish ranges and price each deal individually. Treat the money as one-off.
  • Rights come first. Selling a product built on aggregated data is a different legal exercise from handing over raw records, and client contracts and personal data can rule much of it out.
  • Your options can narrow. An exclusive licence could stop you doing anything else with the same data.

The programmes that buy company records publish ranges rather than prices. See what your data is worth for those figures, and the public deals tracker for what big content owners have been paid. Those headline deals are a different market.

03Licence or sale?

For almost every company, a licence is the better form. You keep the original, and you can set limits.

Licensing versus selling outright: under a licence you keep ownership and grant agreed use of an agreed copy; in an outright sale the rights pass to the buyer. Most AI data programmes license.LicenceOutright saleOwnershipYou keep itPasses to the buyerWhat changes handsPermission to use an agreed copy, foragreed purposesThe data and the rights in itWhat to readUse, exclusivity, term, resale,deletionEverything: you can’t take it back
Most programmes license. The licence terms matter more than the word “sell”.

04Which should you do first?

If you have no plan for your data, MIT's numbers point to starting at home: improving your own work has the best record, and it costs nothing in rights risk. If you already sell software or services, wrapping is usually next.

Licensing doesn't compete with either. It uses records you're already keeping, it needs little of your team's time until a programme is interested, and a non-exclusive licence leaves those other options open. A data inventory is useful for all of them.

For the practical route, see how to get paid for your company's data.

05Questions people ask

What is data monetization?

Turning data into money or measurable value, either directly by selling access to it, or indirectly by using it to cut costs or improve products.

What's the difference between selling data and licensing it?

A sale passes the data and rights to the buyer. A licence gives permission to use an agreed copy for agreed purposes, and you keep the original.

Is licensing to AI a recurring income?

Don't plan on it. The programmes describe individually priced deals, so treat it as a one-off until you have a contract that says otherwise.

Sources

Our links to these programmes carry our referral code, so the programme can see the introduction came from us. The programme may pay us a referral fee, and you pay nothing.

Check which programmes you’re likely to matchA few short questions, about a minute. Free, no sign-up. The programme decides who qualifies.trackimpact.io/data →