# Data Valuation: How to Value Company Data

> Data valuation explained: the cost, market and income methods, what AI buyers price on, a worked example from public deals and the mistakes to avoid.

Canonical: https://origindatapartners.com/guides/data-valuation
By: Origin Data Partners
Status: Published
Published: 2026-11-10
Updated: 2026-11-10
Sources checked: 2026-11-09

## Quick answer

Data valuation puts a money figure on a body of records. Three methods do the work: cost (recreating the records), market (what comparable data fetched) and income (the discounted cash it can bring in). Cost usually marks the floor and comparable deals the ceiling. For an AI license, the buyer's terms decide where you land.

## Key points

- No single standard exists. Use the cost, market and income methods side by side and compare the answers.
- Cost-based numbers tend to be a floor; market-based numbers tend to be a ceiling.
- AI buyers price on scarcity, history, structure, rights and the permitted use, not on volume alone.
- A value is only real once a buyer agrees to terms. Treat every estimate as a range.

## What data valuation means for a company

Valuing data is harder than valuing a machine or a building. The Harvard Data Science Review puts it plainly: "Today, there is no standard to measure the value of data." ([Harvard Data Science Review](https://hdsr.mitpress.mit.edu/pub/1qxkrnig)). A UK government review reached the same view: there is no consensus on the best approach, and most methods fall into cost-based, market-based or use-based groups ([GOV.UK](https://www.gov.uk/government/publications/value-of-public-sector-data-estimate/value-of-public-sector-data-estimate)).

In practice, a company values data for one of three reasons: to set a price before licensing it, to judge an offer it has received, or to account for it in a sale of the business. The method you lean on depends on the reason. For an AI license, what matters is what a buyer would pay for defined rights to a prepared copy.

This guide covers the methods and how to use them. For published prices by record type and what moves them, see [what company data is worth](https://origindatapartners.com/guides/company-data-value). For the accounting view, see [whether data belongs on the balance sheet](https://origindatapartners.com/guides/company-data-value#data-on-the-balance-sheet).


## The three methods: cost, market and income

The same three approaches used for patents and other intangible property apply to data. The US Internal Revenue Manual sets them out for intangibles ([IRS](https://www.irs.gov/irm/part4/irm_04-048-005)), and the Harvard review notes that, as with physical assets, data can be valued on its cost, its sale value or its income potential.

| Method | The question it answers | How to apply it to company data | Main weakness |
| --- | --- | --- | --- |
| Cost | What would it cost to recreate these records? | Add up staff time, systems and years needed to produce the same history; the IRS starts from "the cost to reproduce the intangible" and adjusts historical cost for obsolescence ([IRS](https://www.irs.gov/irm/part4/irm_04-048-005)) | Records produced as a by-product of operations cost little to collect, so the answer tends to be a floor (GOV.UK calls it a lower bound) |
| Market | What has comparable data sold or licensed for? | Find deals for similar records, with similar terms, rights, exclusivity and duration, and adjust for differences | Prices for company data are rarely public, and published deals mix data with other assets, so the answer tends to be a ceiling |
| Income | What future cash will the data earn? | Forecast license fees or savings, then discount them to today, as the IRS describes with net present value | Depends on forecasts of earnings, duration and risk that are hard to check |

Source: [IRS Internal Revenue Manual 4.48.5: Intangible Property Valuation Guidelines](https://www.irs.gov/irm/part4/irm_04-048-005)

## Why the methods give a range, not one number

The UK review explains why the answers differ. Cost-based methods "are likely to provide a lower-bound estimate of the value of data", while market-based methods are likely to provide an upper bound, because it is difficult to separate the value of data from other digital and intangible assets ([GOV.UK](https://www.gov.uk/government/publications/value-of-public-sector-data-estimate/value-of-public-sector-data-estimate)). It also notes that market prices for data are often not available, because data is usually produced and analysed inside the company.

Income methods sit in between and swing with your assumptions. When applied to businesses, the review says, use-based methods are often called "income-based" or "revenue-based" methods. The IRS warns that the income method relies on estimates of future earnings, how long they last and the risk of not realising them ([IRS](https://www.irs.gov/irm/part4/irm_04-048-005)).

So run all three and read them together. If the cost floor is above anything a buyer would pay, licensing is probably not worth the preparation. If comparable deals sit well above the floor, the work is in proving that your records are comparable.


## What AI buyers price on

An AI developer is not buying your storage bill. It is buying the ability to train, test or ground a model on records it cannot get elsewhere. In Origin's experience as an independent data partner, five things move the number most.

- Scarcity: records that never appeared on the public web, such as resolved support cases, internal decisions or operational logs.
- History and continuity: years of dated records with few gaps are worth more than a recent snapshot.
- Structure and links: tickets joined to outcomes, or documents joined to the decisions they led to, beat loose exports.
- Rights: clear authority to license, with customer, employee and third-party material handled. Unclear rights cut the price or end the deal.
- Permitted use and terms: training, evaluation and retrieval are priced differently, and exclusivity, duration and territory change the number. The IRS lists the same terms when it tests whether a deal is comparable ([IRS](https://www.irs.gov/irm/part4/irm_04-048-005)).


## Worked example: valuing a small company archive

The company below is an illustration, not a client. The comparable deal is real and described without names; the [AI data licensing deals tracker](https://origindatapartners.com/guides/ai-data-licensing-deals) lists the parties and sources for every public deal.

Take a software company with eight years of support tickets, linked to product releases and fixes, plus internal chat about the same issues. It is weighing a non-exclusive license to an AI developer for training.

| Method | Inputs | What it suggests |
| --- | --- | --- |
| Cost | Two engineers for six weeks to extract, de-identify and document the records, plus counsel to review rights | A floor: below this, the license does not repay the preparation |
| Market | Reported deals for similar records. MIT Technology Review describes a closed startup that netted roughly $25,000 for its code base, with about $15,000 likely for its workspace data ([MIT Technology Review](https://www.technologyreview.com/2026/10/13/1145755/ai-labs-buying-data-from-dead-startups-to-train-models/)). Licenses for published content run far higher, but rates vary by orders of magnitude by content and buyer ([Quartz](https://qz.com/ai-training-data-pricing-licensing-deals-market-052126)) | A reference range for a small archive, adjusted up for longer history and linked outcomes, down for gaps and rights work |
| Income | Possible renewals or further non-exclusive licenses of the same records, discounted for the chance they do not happen | Upside only if the records stay distinctive and the first license allows reuse |


## Reading the example

The three answers frame a negotiation rather than set a price. The cost floor tells the owner when to walk away. The comparable deal suggests a small archive can trade in the tens of thousands of dollars rather than millions; larger, longer or rarer records can fetch more. The income view shows why a non-exclusive license can be worth more over time than a one-off sale.

Small sellers also face a weak position. MIT Technology Review notes that in shutdown sales "there's little assurance they're getting a fair offer" ([MIT Technology Review](https://www.technologyreview.com/2026/10/13/1145755/ai-labs-buying-data-from-dead-startups-to-train-models/)). Running the three methods before the first call is the best protection.

For an indicative range for your own records, [answer six questions](https://origindatapartners.com/check-fit). Published ranges and estimates are not offers.


## Common data valuation mistakes

Before a buyer makes an offer, use the [offer worksheet](https://origindatapartners.com/guides/company-data-value#offer-worksheet) to compare proposals on more than the headline number. See also [data monetization](https://origindatapartners.com/guides/data-monetization) for the routes to earn from data, and [AI licensing](https://origindatapartners.com/guides/ai-licensing) for how license deals are structured.

- Using a headline deal as a benchmark. The largest publicly reported licenses are for large archives of published content, not for one company's internal records.
- Pricing by volume. Gigabytes and row counts matter less than history, links between records and rights.
- Valuing data you cannot license. If contracts or privacy rules block the use, the value to a buyer is close to zero until that is fixed.
- Ignoring the terms. The same records are worth different amounts for training, evaluation or retrieval, and exclusive rights should cost the buyer more.
- Treating cost as value. What the data cost to collect sets a floor, not a price.
- Skipping a written scope. Without a defined extract, buyer and seller are valuing different things.


## Questions and answers

### How do you value data?

Use three methods side by side: cost (what it would take to recreate the records), market (what comparable data has sold or licensed for) and income (the discounted future cash it can earn). Read the answers as a range: cost tends to be the floor and comparable deals the ceiling.

### What are the three data valuation methods?

Cost, market and income. They are the same approaches the US Internal Revenue Manual uses for other intangible property ([IRS](https://www.irs.gov/irm/part4/irm_04-048-005)). A UK government review groups data methods as cost-based, market-based and use-based, and notes that use-based methods applied to businesses are often called income-based.

### How much is my company data worth?

It depends on what the records are, how far back they go, whether they link to outcomes and whether you can license them. One reported sale of a small startup archive came to tens of thousands of dollars. See [what company data is worth](https://origindatapartners.com/guides/company-data-value) for published prices by record type.

### Is there a standard way to value data?

No. The Harvard Data Science Review says there is no standard to measure the value of data ([Harvard Data Science Review](https://hdsr.mitpress.mit.edu/pub/1qxkrnig)), and a UK government review found no consensus on the best approach. That is why owners compare several methods.

### Which data valuation method should a small company use?

Start with cost to find your floor, then look for comparable deals in the [deals tracker](https://origindatapartners.com/guides/ai-data-licensing-deals). Use income only if you expect repeat licenses. The deciding input is usually what a buyer will pay for a defined use.

### Why do data valuations differ so much?

Because each method answers a different question. A UK government review says cost-based methods tend to give a lower bound and market-based methods an upper bound, partly because market prices for data are rarely available ([GOV.UK](https://www.gov.uk/government/publications/value-of-public-sector-data-estimate/value-of-public-sector-data-estimate)).

## Sources

- [Harvard Data Science Review: a review of data valuation approaches (2023)](https://hdsr.mitpress.mit.edu/pub/1qxkrnig)
- [GOV.UK: Value of public sector data estimate (Oct 2025)](https://www.gov.uk/government/publications/value-of-public-sector-data-estimate/value-of-public-sector-data-estimate)
- [IRS Internal Revenue Manual 4.48.5: Intangible Property Valuation Guidelines](https://www.irs.gov/irm/part4/irm_04-048-005)
- [MIT Technology Review: AI labs buying data from shut-down startups (Oct 2026)](https://www.technologyreview.com/2026/10/13/1145755/ai-labs-buying-data-from-dead-startups-to-train-models/)
- [Quartz: the price of AI training data and licensing deals (2026)](https://qz.com/ai-training-data-pricing-licensing-deals-market-052126)

## Related reading

- [How much do AI companies pay for company data?](https://origindatapartners.com/guides/company-data-value)
- [AI data licensing deals tracker: who paid what, 2023 to 2026](https://origindatapartners.com/guides/ai-data-licensing-deals)
- [Data licensing pricing models: flat fee, recurring and revenue share](https://origindatapartners.com/guides/data-licensing-pricing-models)
- [Data Monetization: Meaning, Strategies and Platforms](https://origindatapartners.com/guides/data-monetization)
- [AI Licensing: How Data and Content Deals With AI Work](https://origindatapartners.com/guides/ai-licensing)