# How to Monetize Data: A Practical Guide for Companies

> How a company can monetize data: internal use, data products, insights or licensing de-identified data to AI labs, compared on cost, risk and time to cash.

Source: https://sellmydata.com/guides/how-to-monetize-data/
Published: 2026-10-28 · Updated: 2026-10-28

## Key takeaways

- A company can monetize data indirectly, by using it to cut costs or earn more, or directly, by selling data, information products or services. Gartner's definition, as quoted by one analytics vendor, calls it using data to obtain quantifiable economic benefit.
- Most of the value often comes from using data, not selling it: MIT researchers found that, on average, 82% of data monetization returns come from improving work operations and customer experience, versus 18% from selling information.
- Licensing de-identified business data to AI labs for model training is a newer direct route that does not require building a product. Reported deals vary widely and are not benchmarks.
- Whatever the route, three questions decide value and feasibility: what data you hold, whether you have the rights to use it that way, and whether the people in it can be reliably de-identified.

A company can monetize its data in two broad ways: indirectly, by using it to make better decisions and cut costs, or directly, by selling or licensing data, insights or data products. For a small or mid-size company, the practical options run from improving internal operations to licensing de-identified business data to AI labs for model training. This guide maps each route, compares them on effort, cost, risk and time to cash, and ends with a short checklist.

## What is data monetization?

[Data monetization](/glossary/data-monetization/) (or data monetisation, in British spelling) means turning data into measurable economic value. Gartner's definition, as quoted in [one analytics vendor's explainer](https://www.gooddata.ai/blog/data-monetization-8-things-you-should-know/), is "the process of using data to obtain quantifiable economic benefit." MIT's Center for Information Systems Research (CISR) describes it more simply as [converting data into money](https://cisr.mit.edu/publication/data-everybodys-business-book).

The same Gartner definition, [as quoted by that vendor](https://www.gooddata.ai/blog/data-monetization-8-things-you-should-know/), splits the methods in two:

- **Indirect (internal) monetization:** using data to make measurable business improvements and inform decisions. Example: a support team studies years of tickets, finds the top reasons customers cancel, and fixes them.
- **Direct monetization:** sharing data to gain better terms from partners, information bartering, selling data outright, or offering information products and services. Example: licensing a de-identified archive of those same support tickets to an outside buyer.

MIT CISR names [three ways to create bottom-line impact from data](https://cisr.mit.edu/publication/2015_1201_DataDeluge_WixomRoss): improving internal processes and decisions, wrapping information around core products and services, and selling information offerings. Its researchers also argue that most organizations [view data monetization too narrowly](https://cisr.mit.edu/publication/data-everybodys-business-book), as merely selling data sets. Keeping all three in view leads to better choices.

**Scope of this guide.** This is about business data held by companies: messages, documents, code, tickets and records. It is not about individuals selling their own personal information. For reference, SellMyData, the exchange that publishes this guide ("we" below), [does not buy](https://sellmydata.com/sell-my-data/) consumer [personal data](/glossary/pii/), health records tied to people, or payment card data.

## Indirect data monetization: using data to earn or save more

Indirect data monetization means using your own data inside the company to raise revenue or lower costs, without the data leaving the building. MIT research suggests most returns come from improving operations and customer experience rather than from selling data.

MIT CISR calls the internal approach **improving**, which it defines as [optimizing internal processes and decisions](https://cisr.mit.edu/publication/2025_0201_DataMonetizationImpact_vanderMeulenWixomBeath). Typical uses include:

- **Pricing:** using deal history to see which discounts actually win business.
- **Forecasting:** using past orders and tickets to plan staffing and inventory.
- **Churn reduction:** spotting the patterns that come before a customer leaves.
- **Operational efficiency:** finding where work gets stuck in approvals, handoffs or support queues.

For context: MIT researchers found that, on average, [82% of data monetization returns come from improving work operations and customer experience](https://cisr.mit.edu/sites/default/files/2023-10/Data%20is%20Everybody%27s%20Business%20Press%20Release%20101623.pdf), versus 18% from selling information. That figure covers both internal improvements and customer-facing enhancements (what MIT calls wrapping, covered below).

**Pros:** the data does not leave the company, so there is no outside buyer to vet and fewer contract questions. Internal use of personal data can still be regulated, so check with counsel.

**Cons:** value can be hard to measure and slow to show up. MIT CISR's 2023 briefing makes the point that efficiency or customer value only counts as monetization once it is [converted into money](https://cisr.mit.edu/publication/2023_0801_DataMonetization_WixomBeathOwens), for example as lower costs or higher sales. If no one tracks that conversion, the gain stays invisible.

## Direct data monetization: four routes compared

Direct monetization brings in revenue from the data itself or from things built on it. For a smaller company, the four realistic routes are data products, insights services, wrapping data into existing products, and licensing de-identified datasets. They differ most in how much you must build before the first dollar arrives.

### Data products and APIs

An API (application programming interface) is a way for other software to request your data automatically. Building a data product or API means packaging data so customers can pay for ongoing access. MIT CISR calls this kind of route **selling**: [developing new revenue-generating information solutions](https://cisr.mit.edu/publication/2025_0201_DataMonetizationImpact_vanderMeulenWixomBeath). It can produce recurring revenue, but it needs engineering to build and maintain, plus a sales effort to find customers.

### Insights and analytics services

Instead of handing over data, you sell conclusions drawn from it: reports, benchmarks or advisory work. This needs analysts who can turn raw data into something a customer will pay for, and people to sell it.

### Wrapping data into existing products

MIT CISR defines **wrapping** as [delivering a product enhancement that customers value](https://cisr.mit.edu/publication/2025_0201_DataMonetizationImpact_vanderMeulenWixomBeath). It can take the form of a [dashboard, report, alert, benchmark, suggested next step, or automated action](https://cisr.mit.edu/publication/2018_1201_WrappingValue_WixomSchuritz) inside a product you already sell, sometimes as a premium tier. You keep the customer relationship you have, but you still need product and engineering time.

### Licensing de-identified datasets

[De-identification](/glossary/de-identification/) means removing or replacing the details that identify people. Licensing means granting a buyer permission to use a dataset under agreed terms. This route, which now includes licensing to AI labs for model training, needs little product build. We [price datasets on four drivers](https://sellmydata.com/pricing/#how-we-estimate): data type, volume, years of history and current lab demand.

### How the routes compare

| | Data products and APIs | Insights services | Wrapping into products | Licensing de-identified data |
|---|---|---|---|---|
| Upfront cost | High (engineering build) | Medium (analyst time) | Medium (product work) | Low (export, cleaning, review) |
| Ongoing effort | High: maintenance, support, sales | High: analysis and selling | Medium: feature upkeep | Low after the dataset is prepared |
| Time to first revenue | Slow: build first, then sell | Moderate: depends on landing clients | Moderate: tied to product releases | Depends on finding a buyer and agreeing terms |
| Control over the data | You host it; customers get access | Data stays in-house; only conclusions leave | Data stays inside your product | Buyer gets a copy under contract terms |
| Main risk | Build cost without enough customers | Hard to scale beyond people's time | Customers may not pay more | Rights, privacy and re-identification |

None of these is right for every company. If you have no engineering capacity, data products are a stretch. If your data mostly describes your own operations, indirect use may pay back more than any sale.

## Why AI training created a new way to monetize data assets

AI labs need real-world material to train models, and some have paid for business data such as emails, chat messages, code and tickets. That gives companies a way to monetize data assets without building a product first. For a fuller explanation, see [what AI data licensing is](/guides/what-is-ai-data-licensing/).

Fast Company reports that advanced agentic models (AI systems that can make decisions on their own) [require more intricate datasets](https://www.fastcompany.com/91528808/shuttered-startups-are-selling-old-slack-chats-and-emails-to-ai-companies), including documents, emails and FAQs that provide context and feedback. It has also noted that big AI labs [have the money to license or buy](https://www.fastcompany.com/91627173/why-you-should-worry-less-about-big-data-centers-and-more-about-big-data) massive amounts of business and consumer data. For our part, we list only de-identified business data and do not buy consumer personal data.

Publicly reported deals give a sense of range, though each is a single case, not a benchmark:

- Reuters reported in February 2024 that Reddit's content-licensing deal with Google for AI training is worth [about $60 million per year](https://tech.yahoo.com/ai/articles/exclusive-reddit-ai-content-licensing-041012503.html), according to a source. Reddit is a content platform licensing user-generated posts, which is different from a company licensing its own internal business data.
- Axios reported in August 2026 that Google [won a bankruptcy auction for Spirit Airlines' business data](https://www.axios.com/2026/08/17/google-spirit-airlines-bankruptcy) with a $10 million bid, subject to court approval. A court filing says the data includes about 100 million emails and 500 million Microsoft Teams chats, which Google will use to improve its products and AI models. Bloomberg Law describes it as [de-identified business data, software code and operations records](https://news.bloomberglaw.com/bankruptcy-law/google-aims-to-boost-ai-with-purchase-of-spirit-airlines-data).
- Fast Company, citing Forbes, reported that the CEO of one defunct software company sold every Slack message, internal email and Jira ticket as [AI training data](/glossary/ai-training-data/) for ["hundreds of thousands of dollars"](https://www.fastcompany.com/91528808/shuttered-startups-are-selling-old-slack-chats-and-emails-to-ai-companies).

### Licensing versus selling outright

In general terms, a sale transfers the data while a license grants permission to use it under conditions; what actually applies depends on the agreement. Conditions a license can set include scope (what is included and for what use), term (how long) and [exclusivity](/glossary/exclusive-license/) (whether you can license the same data to others); on our exchange, scope, format, exclusivity and term are agreed per dataset. Conditions can attach even to a sale: the Spirit sale [excludes passenger profiles and frequent-flyer information](https://finance.yahoo.com/technology/ai/articles/google-bids-10-million-spirit-150157527.html), an independent third party will scrub the material before it is transferred, and court filings state the buyer may not attempt to identify individuals. This is general information; have counsel review any agreement before you sign.

## How to build a data monetization strategy in five steps

A data monetization strategy is a plan for which data you will turn into value, by which route, under what rights and safeguards. For a smaller company it fits on a page and follows five steps.

### 1. Inventory your data

List the tools you use, the date ranges they cover and rough volumes. Start with where messages and documents live:

- [How to check Slack data volume](/guides/how-to-check-slack-data-volume/)
- [How to check Google Drive data volume](/guides/how-to-check-google-drive-data-volume/)
- [How to check Microsoft Teams data volume](/guides/how-to-check-microsoft-teams-data-volume/)
- [How to check OneDrive and SharePoint data volume](/guides/how-to-check-onedrive-sharepoint-data-volume/)

Add code repositories, ticketing systems and record-keeping tools. Our [copy-and-paste inventory prompt](https://sellmydata.com/inventory/) can help you list them.

### 2. Check your rights

Before any external use, review four areas: customer contracts, your privacy policy, employee communications and third-party content (material created by someone else, such as vendor documents). Whether you may use data a certain way depends on what those agreements and policies say and on the laws that apply to you. This is general information, not legal advice; ask counsel to confirm. For a checklist of what we review with companies before listing, see our [rights page](https://sellmydata.com/rights/).

### 3. Plan de-identification

One approach is to remove names, contact details and identifiers and replace them with consistent placeholders, so a conversation still reads naturally without revealing who said what. Placeholders alone are not the whole job: free text can carry identifying details, which is why human review matters. Read [how to anonymize data for AI](/guides/how-to-anonymize-data-for-ai/) for methods and limits, and see our [de-identification page](https://sellmydata.com/de-identification/) for what we remove and where the limits are.

### 4. Estimate value

We do not publish a price list, and the deals above are not benchmarks. We base estimates on four drivers: data type, volume, years of history and current lab demand. [Our explainer](https://sellmydata.com/pricing/#how-we-estimate) shows how those turn into an indicative range.

### 5. Pick a route, run a pilot, keep the ability to stop

Choose one route and test it small: one indirect project with a clear metric, or one dataset from one tool and date range. Make sure you can stop, whether that means ending an internal project or, for a license, having terms that let you unlist and delete. What those terms allow depends on the agreement; have counsel check them.

## Risks and how to manage them

The main risks are privacy and re-identification, contract and consent limits, and reputation with customers and staff. Each can be managed with preparation, and indirect monetization keeps data inside the company, though internal use can still be regulated.

- **Re-identification.** Removing names does not always make data anonymous. Research summarized by Imperial College London found that [99.98% of Americans would be correctly re-identified](https://www.imperial.ac.uk/news/192112/anonymising-personal-data-enough-protect-privacy/) in any dataset using 15 demographic attributes. The lesson is not to give up, but to treat de-identification as careful work with human review.
- **Contract and consent limits.** An FTC staff post (February 2024, as quoted by a privacy law blog) warned that it ["may be unfair or deceptive"](https://www.theprivacylaw.com/2024/02/ftc-says-if-you-want-to-use-data-in-a-new-way-you-need-to-provide-actual-notice-to-end-users-before-you-do-it/) for a company to start using consumers' data for AI training and only inform them through a surreptitious, retroactive change to its terms of service or privacy policy. The FTC also says that in past enforcement it has required businesses that unlawfully obtained consumer data to [delete products, including models and algorithms](https://www.ftc.gov/policy/advocacy-research/tech-at-ftc/2024/04/consumer-facing-applications-quote-book-tech-summit-ai), built with it. This is general information; ask counsel how it applies to you.
- **Reputation.** Staff and customers care how their communications are used. In the Spirit sale, the flight attendants' union [objected and sought additional protections](https://finance.yahoo.com/technology/ai/articles/google-bids-10-million-spirit-150157527.html) for employee data.

### What the rules say about any dataset (general information)

These points apply to any dataset, not to a specific provider, and are general information rather than legal advice.

- GDPR Recital 26 says data protection principles [should not apply to anonymous information](https://gdpr-info.eu/recitals/no-26/): information that does not relate to an identifiable person, or personal data rendered anonymous so the person is no longer identifiable. This is what [anonymization](/glossary/anonymization/) aims for.
- The same recital says pseudonymised data (data where identifiers are swapped out but could be linked back to a person using extra information) [should still be treated as information on an identifiable person](https://gdpr-info.eu/recitals/no-26/).
- The UK ICO says pseudonymisation is [effectively only a security measure](https://ico.org.uk/for-organisations/uk-gdpr-guidance-and-resources/personal-information-what-is-it/what-is-personal-data/what-is-personal-data/) and warns that organisations frequently call data sets anonymised when they are not.

Which standard applies to your data depends on where you and the people in it are located and on your agreements. Ask counsel to confirm before any external use.

### How to manage the risks

- De-identify thoroughly and have a person review the output.
- Run a rights review covering contracts, privacy policy, employee communications and third-party content.
- Approve each buyer yourself rather than allowing open resale.
- Let buyers see only a redacted sample before purchase.
- Keep the ability to unlist and delete.
- Tell staff and, where relevant, customers what you plan, in plain language.

## Where a managed exchange fits

If licensing looks like the right route, an exchange can handle the preparation work so you do not have to build anything. SellMyData is one such option for small and mid-size companies.

There is no minimum headcount and no sales call to list. You answer a few questions to get an estimated price range before sharing any data, and a person reviews every request and replies by email. You export data from the tools and date ranges you choose; we clean and prepare it, replace names, contact details and identifiers with consistent placeholders, run a rights check with you, and you approve the result. We never write to a company's systems.

Buyers see only a redacted sample before purchase, and a licensed lab receives the de-identified dataset, never raw data. Nothing is licensed without the seller's sign-off on the buyer, the terms and the price, and sellers can [unlist and delete their data at any time](https://sellmydata.com/trust/). There is nothing to pay up front: we take a commission only when a sale closes, with a target of payment within 7 days of a sale closing ([pricing](https://sellmydata.com/pricing/)). For the full process, read [how to sell company data to AI labs](/guides/how-to-sell-company-data-to-ai-labs/).

To see where you stand, use the quick estimate on the [homepage](https://sellmydata.com/) to get an indicative range. The range assumes you keep your full history and license it exclusively, is calibrated to reported deals and client quotes as of October 2026, and is not an offer.

And if your inventory suggests the bigger payoff is internal, start there. Licensing is one way to monetize data, not the answer for everyone.

## Frequently asked questions

### What is data monetization?

Data monetization is turning data into measurable economic value. Gartner's definition, as quoted by one analytics vendor, is the process of using data to obtain quantifiable economic benefit, either indirectly through better decisions or directly through sharing, bartering, selling data or offering information products.

### How can you monetize data?

You can use it internally to improve pricing, forecasting and efficiency, wrap it into your products as premium features, sell insights or data products, or license de-identified datasets, including to AI labs for training. Each route differs in upfront cost, effort, risk and time to revenue.

### How do you monetize data assets?

Start by inventorying what you hold by tool, date range and volume, then check your rights to use it, plan de-identification and estimate its value. Then pick one route, run a small pilot and keep the ability to stop.

### What is a data monetization strategy?

A data monetization strategy is a plan for which data you will turn into value, by which route, under what rights and safeguards, and how you will measure the result. For a smaller company it can be as simple as one inventory, one rights review and one pilot.

### What is indirect data monetization?

Indirect data monetization means using data inside the company to make measurable improvements, such as better pricing, lower churn or more efficient operations, rather than selling it. The data does not leave the company, but the value can be harder to measure and slower to appear.

### Is selling data profitable for a small company?

It can be, but there is no guaranteed figure. Reported examples include one defunct software company's sale for hundreds of thousands of dollars and Google's $10 million winning bid, subject to court approval, for Spirit Airlines' business data; Reddit's reported deal with Google (about $60 million a year, per a Reuters source) involves a content platform, which differs from a company licensing internal data. These are individual reports, not benchmarks; at SellMyData, price depends on data type, volume, years of history and current lab demand.

### How can I monetize my data if I'm an individual rather than a company?

This guide covers business data held by companies, not individuals selling their own personal information. SellMyData does not buy personal data from individuals, so read the terms of any consumer program carefully before sharing anything.

### Can a company monetize data without sharing personal information?

Often, yes, with care. Indirect monetization keeps data inside the company, and direct licensing can be done with datasets where names, contact details and identifiers have been removed and checked by a person. Because re-identification is a real risk and swapping names for placeholders alone may not make data anonymous in law, counsel should confirm how the result is treated under the laws that apply to you.

### What is the difference between licensing and selling data?

In general terms, a sale transfers the data, while a license grants permission to use it under conditions such as scope, term and exclusivity. What actually applies depends on the agreement, so have counsel review any agreement before you sign.

### What does wrapping mean in data monetization?

MIT CISR defines wrapping as delivering a product enhancement that customers value, such as a dashboard, report, alert, benchmark, suggested next step or automated action inside a product you already sell. You keep your existing customer relationship, but it still takes product and engineering time.

### How is a de-identified dataset priced for AI training?

SellMyData prices datasets on four drivers: data type, volume, years of history and current lab demand.

## Sources

- [Data Monetization: 8 Things You Should Know](https://www.gooddata.ai/blog/data-monetization-8-things-you-should-know/)
- [The Data Deluge (Wixom & Ross, 2015)](https://cisr.mit.edu/publication/2015_1201_DataDeluge_WixomRoss)
- [Data Monetization Impact (van der Meulen, Wixom & Beath, 2025)](https://cisr.mit.edu/publication/2025_0201_DataMonetizationImpact_vanderMeulenWixomBeath)
- [Data Monetization research briefing (Wixom, Beath & Owens, 2023)](https://cisr.mit.edu/publication/2023_0801_DataMonetization_WixomBeathOwens)
- [Wrapping Value (Wixom & Schüritz, 2018)](https://cisr.mit.edu/publication/2018_1201_WrappingValue_WixomSchuritz)
- [Data Is Everybody's Business (book)](https://cisr.mit.edu/publication/data-everybodys-business-book)
- [Data Is Everybody's Business press release (October 2023)](https://cisr.mit.edu/sites/default/files/2023-10/Data%20is%20Everybody%27s%20Business%20Press%20Release%20101623.pdf)
- [Exclusive: Reddit in AI content licensing deal with Google](https://tech.yahoo.com/ai/articles/exclusive-reddit-ai-content-licensing-041012503.html)
- [Google wins Spirit Airlines bankruptcy auction for business data](https://www.axios.com/2026/08/17/google-spirit-airlines-bankruptcy)
- [Google Aims to Boost AI With Purchase of Spirit Airlines Data](https://news.bloomberglaw.com/bankruptcy-law/google-aims-to-boost-ai-with-purchase-of-spirit-airlines-data)
- [Google bids $10 million for Spirit Airlines data](https://finance.yahoo.com/technology/ai/articles/google-bids-10-million-spirit-150157527.html)
- [Shuttered startups are selling old Slack chats and emails to AI companies](https://www.fastcompany.com/91528808/shuttered-startups-are-selling-old-slack-chats-and-emails-to-ai-companies)
- [Why you should worry less about big data centers and more about big data](https://www.fastcompany.com/91627173/why-you-should-worry-less-about-big-data-centers-and-more-about-big-data)
- [GDPR Recital 26: Not Applicable to Anonymous Data](https://gdpr-info.eu/recitals/no-26/)
- [What is personal data?](https://ico.org.uk/for-organisations/uk-gdpr-guidance-and-resources/personal-information-what-is-it/what-is-personal-data/what-is-personal-data/)
- [FTC says if you want to use data in a new way, you need to provide actual notice to end users before you do it](https://www.theprivacylaw.com/2024/02/ftc-says-if-you-want-to-use-data-in-a-new-way-you-need-to-provide-actual-notice-to-end-users-before-you-do-it/)
- [Consumer Facing Applications: A Quote Book from the Tech Summit on AI](https://www.ftc.gov/policy/advocacy-research/tech-at-ftc/2024/04/consumer-facing-applications-quote-book-tech-summit-ai)
- [Anonymising personal data 'not enough to protect privacy'](https://www.imperial.ac.uk/news/192112/anonymising-personal-data-enough-protect-privacy/)
- [SellMyData homepage and estimate](https://sellmydata.com/)
- [Can I sell my data?](https://sellmydata.com/sell-my-data/)
- [Rights check](https://sellmydata.com/rights/)
- [De-identification](https://sellmydata.com/de-identification/)
- [Pricing: how we estimate](https://sellmydata.com/pricing/#how-we-estimate)
- [How sellers stay in control](https://sellmydata.com/trust/)
- [Data inventory prompt](https://sellmydata.com/inventory/)