How much is my company's data worth?

Short answer: Your company's data is worth what a willing buyer would pay to use it, which usually means licensing access rather than selling it outright. Most companies cannot see this value in their accounts, because accounting rules generally do not let a business record data it generated itself as an asset. Value depends on eight drivers: volume, history, freshness, uniqueness, rights and consent, quality, regulatory exposure and buyer demand. Valuers estimate it with the three approaches used for intangible assets: cost, income and market. There is no reliable price per record, but an indicative range takes about five minutes to estimate.

Last updated October 10, 2026. Published by Data Revenue.

Key takeaways

• Data is an intangible asset. Its value comes from what a buyer can do with it, not from what it cost you to collect.
• Data a company generates itself rarely appears on its balance sheet under IFRS or US GAAP, so book value tells you almost nothing about market value.
• Real transactions show that data-rich assets can carry large values. Reddit disclosed $203.0 million of data licensing contracts in its IPO filing, and United Airlines valued its MileagePlus loyalty program at about $21.9 billion when it borrowed against it.
• Those figures come from very large, unusual assets. They are reference points, not benchmarks for a typical company.
• The same dataset can be worth very different amounts to different buyers, depending on eight value drivers.
• Professional valuations combine the cost, income and market approaches, following IVS 210 methodology.

Why your data is probably not on your balance sheet

A common first question is: "If our data is so valuable, why isn't it in our accounts?" The answer is accounting rules, not a lack of value.

Under IFRS (IAS 38)

IAS 38 Intangible Assets governs intangible assets for companies reporting under IFRS. Paragraph 63 contains an explicit prohibition:

"Internally generated brands, mastheads, publishing titles, customer lists and items similar in substance shall not be recognised as intangible assets." (IAS 38, paragraph 63, as quoted in an IASB and FASB staff paper, IFRS Foundation, October 11, 2024)

The same paper summarises that intangible assets can be carried at fair value under the revaluation model only where fair value is measured by reference to an active market, which most datasets do not have. It also lists, as an open question for the IASB's research project, whether there should be a recognition difference between acquired and internally generated intangible assets (IFRS Foundation, October 11, 2024).

Under US GAAP (ASC 350-30)

US GAAP reaches a similar result. ASC 350-30-25-3 requires that costs of internally developing, maintaining or restoring intangible assets that are not specifically identifiable be recognized as an expense when incurred (Deloitte Roadmap: Goodwill and Intangible Assets, section 4.1, accessed October 2026).

What this means in practice

• Data you buy, or acquire as part of buying a company, can generally be recognized. Data you build generally cannot.
• Missing from the balance sheet does not mean worthless. It means the value has to be estimated through a valuation rather than read from the accounts.
• Buyers, lenders and acquirers look at what the data can earn, not at its book value.

What buyers and lenders have actually paid

Most data deals are private. The public examples below are the best verified reference points available, and almost all involve very large companies.

AI training and content licensing

• Reddit: In its IPO registration statement, Reddit disclosed that in January 2024 it entered into data licensing arrangements with an aggregate contract value of $203.0 million and terms of two to three years, and expected at least $66.4 million of that revenue to be recognized in 2024 (TechCrunch, February 22, 2024; Reddit, Inc. Form S-1, February 22, 2024).
• Reddit and Google: Reuters reported that Reddit's content licensing contract with Google is worth about $60 million per year, according to one source (Reuters via Yahoo, February 22, 2024).
• Why Reddit's data commands a premium: Reddit's own filing put it plainly: "Our massive corpus of conversational data and knowledge is what makes us unique" (Reddit, Inc. Form S-1, February 22, 2024).
• Reddit and OpenAI: OpenAI and Reddit announced a partnership giving OpenAI access to Reddit's Data API. Financial terms were not disclosed (OpenAI, May 16, 2024).
• News Corp and OpenAI: According to The Wall Street Journal, News Corp's content deal with OpenAI could be worth more than $250 million over five years (Variety via Yahoo, May 22, 2024).
• Shutterstock: Shutterstock reported $137.3 million of "Data, Distribution, and Services" revenue in 2023, out of $874.6 million total. The category includes licensing metadata and content for AI model training, alongside its Giphy and Studios businesses (Shutterstock Form 10-K, fiscal 2023).

Data-rich assets used as loan collateral

Airline loyalty programs are not pure data assets. They combine member relationships, partner contracts and a very large member database. They are still the clearest public evidence that lenders will lend against a data-rich intangible asset.

• United MileagePlus: In a June 2020 filing, United wrote: "Multiplying MPH 2019 EBITDA by a factor of 12 equates to a MileagePlus valuation of approximately $21.9 billion" (United Airlines Holdings Form 8-K, June 15, 2020). United then raised $6.8 billion ($3.8 billion of senior secured notes and a $3.0 billion term loan) secured by the program (United Airlines Holdings Form 8-K, June 26, 2020).
• Compared with the airline itself: Sherwood News noted that while MileagePlus was valued at close to $22 billion in 2020 bond documents, United's total equity value was $10.6 billion (Sherwood News, August 8, 2024).
• American AAdvantage: American disclosed that a third-party appraisal estimated the AAdvantage program's value at between $19.5 billion and $31.5 billion (American Airlines Group Form 8-K, June 12, 2020). In March 2021 it announced $10.0 billion of AAdvantage financing through notes and a term loan (American Airlines Group, March 10, 2021). American, Delta and United all pledged their loyalty programs as collateral for bonds during the pandemic (Fortune and Bloomberg, May 19, 2021).

How to read these numbers: They show that data-rich assets can be valued and financed at scale. They do not tell you what your data is worth. A mid-sized company's dataset can be worth far less, or occasionally more per record, depending on the drivers below.

The eight drivers of data value

Every serious valuation starts by scoring the data against these factors. Each one can move value up or down significantly.

1. Volume

More records, more entities and more attributes per record generally mean more analytical and training value. Volume alone is rarely decisive: a very large but generic dataset can be worth less than a small, rare one.

2. History and depth

Years of consistent history let buyers study trends, seasonality and outcomes over time. Long time series are especially valued for forecasting, risk modelling and AI training.

3. Freshness

How often the data updates matters. A live or daily feed supports recurring subscription revenue. A static snapshot is usually a one-time sale with a lower ceiling.

4. Uniqueness and exclusivity

The most important driver. If a buyer can get equivalent data elsewhere, your price is capped by the alternative. Data that only your business generates, such as proprietary transaction, sensor, outcome or conversation data, carries a premium. Exclusive licenses usually cost buyers more than non-exclusive ones.

5. Rights and consent

You can only license what you have the right to license. Privacy notices, customer contracts, vendor terms and consent records determine which uses are allowed. Unclear rights are one of the most common reasons value is discounted or a deal cannot proceed.

6. Quality

Completeness, accuracy, consistent definitions, documentation and clean identifiers all raise value, because they reduce the buyer's cost to use the data.

7. Regulatory exposure

Health, financial, children's and location data carry heavier obligations under laws such as HIPAA, GDPR and state privacy laws. Higher regulatory exposure raises the cost of de-identification and compliance, which lowers net value, although well-governed regulated data can still be very valuable.

8. Buyer demand

Value only exists if someone wants to pay. Demand currently comes from AI developers, investment firms, insurers, researchers, marketers and companies in adjacent industries. The more buyer types a dataset serves, the more licenses it can support.

The three valuation approaches (IVS 210 methodology)

The International Valuation Standards Council's standards (effective 31 January 2025) set out how intangible assets are valued using three principal approaches: cost, income and market (IVSC, International Valuation Standards, effective January 31, 2025). Data Revenue's valuations follow IVS 210 methodology. Here is what each approach means for data.

Cost approach

Question it answers: What would it cost to recreate or replace this data with something of similar utility?

The standards describe the cost approach for intangible assets as value based on the replacement cost of a similar asset providing similar utility. For data, that means the cost to collect, clean and accumulate an equivalent dataset, including the years it would take. It is useful as a floor, but it ignores what the data can earn, so it often understates the value of unique data.

Income approach

Question it answers: What income, cash flows or cost savings will this data produce over its useful life, in today's money?

The standards describe this as the most common approach for intangible assets. It values the asset by reference to the present value of income, cash flows or cost savings attributable to it over its economic life. Methods named in the standards include relief-from-royalty, excess earnings, with-and-without, greenfield, distributor and cost savings. For data, relief-from-royalty is a natural fit, because it asks what royalty or license fee a user would pay for access.

Market approach

Question it answers: What have comparable assets sold or licensed for?

The standards allow the market approach for intangible assets only when there is information on arm's-length transactions involving identical or similar assets near the valuation date, plus enough detail to adjust for differences. Because intangible assets are rarely traded separately, such evidence is limited, and the guideline transactions method is generally the only market method available. Public data deals like Reddit's are useful reference points but need large adjustments before they apply to anyone else.

How the approaches fit together

A credible valuation usually considers all three and explains why one is weighted most.

Why there is no reliable "price per record"

You will see per-record figures quoted online. We do not publish a range here, because we could not find a credible, verifiable source that supports one across industries. Values vary so widely because:

• The unit of value is the use case, not the record. The same 1 million records can feed a marketing list, a risk model or an AI training run, each priced differently.
• Exclusivity changes the price. Buyers pay more to keep data away from competitors, so exclusive and non-exclusive licenses of the same data are priced very differently.
• Most deal terms are confidential. Even headline AI deals, such as OpenAI's partnerships with Reddit and with The Associated Press, were announced without financial terms (OpenAI, May 16, 2024; PBS NewsHour, July 13, 2023).
• Rights and regulation can cut value to zero. Data you cannot lawfully license has no external value, however large it is.

The practical answer is to estimate a range for your specific data. You can see an indicative value range in about five minutes with a free 15-question Data Snapshot.

How to get an indicative value for your data

1. List your main data sources. Transactions, customers, products, operations, sensors, content and support interactions.
2. Note the basics for each. Approximate record counts, years of history, update frequency and industry.
3. Check your rights. What do your privacy notices and customer contracts allow?
4. Get an indicative range. The free 15-question Data Snapshot uses those inputs to show an indicative value range in about five minutes.
5. Decide whether a full valuation is worth it. A full valuation adds data profiling, a rights review and documented cost, income and market analysis.

Any figure produced this way is indicative. It is not a guarantee of value or of licensing revenue, and actual results depend on buyers, terms and compliance.

Frequently asked questions

How much is customer data worth per record?

There is no reliable universal price per record. Value depends on uniqueness, history, freshness, rights and consent, quality, regulatory exposure and who wants to buy it. The same records can be worth very different amounts in different use cases, and most deal terms are confidential, so a range estimated for your specific data is far more useful than a generic per-record figure.

Can I put my company's data on the balance sheet?

Usually not, if your company generated the data itself. IAS 38 prohibits recognising internally generated customer lists and similar items as intangible assets, and under US GAAP the costs of internally developing intangible assets that are not specifically identifiable are expensed. Data acquired from others, including through an acquisition, can generally be recognised. The data can still have real economic value for licensing, transactions and financing even when it is not on the balance sheet.

Which valuation approach is best for data?

Most data valuations rely mainly on the income approach, which the International Valuation Standards describe as the most common approach for intangible assets. The cost approach sets a useful floor, and the market approach is used when comparable transactions exist. A credible valuation considers all three and explains how they are weighted.

Does monetizing data mean selling my customers' personal information?

No. The usual model is licensing de-identified, aggregated data products under contracts that prohibit re-identification, while the company keeps ownership of the underlying data. What can be licensed depends on your privacy notices, contracts and the laws that apply to your data.

Why are AI companies paying for data?

AI developers need large volumes of high-quality, lawfully licensed data, and unique human-generated data is hard to replace. Public examples include Reddit's data licensing contracts with an aggregate value of $203.0 million disclosed in 2024 and OpenAI's licensing deals with publishers and content libraries.

How long does it take to value company data?

An indicative range takes about five minutes using a free 15-question Data Snapshot at datarevenue.io/qualify. A full valuation takes longer because it includes data profiling, a review of rights and consent, and documented cost, income and market analysis.

Sources

1. IFRS Foundation (IASB and FASB staff paper, Agenda reference 17B), "Intangible Assets: Update on the project," October 11, 2024. https://www.ifrs.org/content/dam/ifrs/meetings/2024/october/fasb-iasb/ap17b-intangible-assets.pdf
2. Deloitte, "Roadmap: Goodwill and Intangible Assets," section 4.1 (quoting ASC 350-30-25-3), accessed October 2026. https://dart.deloitte.com/USDART/home/codification/assets/asc350-20/goodwill/chapter-4-subsequent-accounting-for-intangible/4-1-overall-accounting-for-intangible
3. Reddit, Inc., Form S-1 Registration Statement, February 22, 2024. https://www.sec.gov/Archives/edgar/data/1713445/000162828024006294/reddits-1q423.htm
4. TechCrunch, "Reddit says it's made $203M so far licensing its data," February 22, 2024. https://techcrunch.com/2024/02/22/reddit-says-its-made-203m-so-far-licensing-its-data
5. Reuters (republished by Yahoo), "Exclusive: Reddit in AI content licensing deal with Google, sources say," February 22, 2024. https://tech.yahoo.com/ai/articles/exclusive-reddit-ai-content-licensing-041012503.html
6. OpenAI, "OpenAI and Reddit Partnership," May 16, 2024. https://openai.com/index/openai-and-reddit-partnership/
7. Variety (republished by Yahoo), "News Corp Inks OpenAI Licensing Deal Potentially Worth More Than $250 Million," May 22, 2024. https://malaysia.news.yahoo.com/news-corp-inks-openai-licensing-211417483.html
8. Shutterstock, Inc., Form 10-K for fiscal year 2023. https://www.sec.gov/Archives/edgar/data/1549346/000154934624000007/sstk-20231231.htm
9. United Airlines Holdings, Inc., Form 8-K (event date June 12, 2020), filed June 15, 2020. https://www.sec.gov/Archives/edgar/data/100517/000110465920073190/tm2022354d3_8k.htm
10. United Airlines Holdings, Inc., Form 8-K (event date June 25, 2020), June 26, 2020. https://www.sec.gov/Archives/edgar/data/100517/000110465920077050/tm2023500d1_8k.htm
11. Sherwood News, "Airline loyalty programs are their most valuable collateral," August 8, 2024. https://sherwood.news/business/jetblue-airline-loyalty-programs-valuable-collateral/
12. American Airlines Group Inc., Form 8-K, June 12, 2020. https://www.sec.gov/Archives/edgar/data/0000006201/000000620120000079/a8k061220operationalan.htm
13. American Airlines Group Inc., Exhibit 99.1 (pricing of AAdvantage notes and term loan), March 10, 2021. https://www.sec.gov/Archives/edgar/data/4515/000000620121000024/aex991securednotespricingp.htm
14. Fortune (Bloomberg), "Travelers be warned: your air miles may be at risk," May 19, 2021. https://fortune.com/2021/05/19/airline-travel-aviation-air-miles-2021
15. International Valuation Standards Council, International Valuation Standards, effective January 31, 2025 (intangible assets standard and valuation approaches; copy hosted by ICAEW). https://icaew.com/-/media/corporate/files/technical/corporate-finance/valuation/ivs-effective-31-january-2025-redline-edition.ashx
16. PBS NewsHour (Associated Press), "ChatGPT-maker OpenAI signs deal with AP to license news stories," July 13, 2023. https://www.pbs.org/newshour/economy/chatgpt-creator-openai-signs-deal-with-ap-to-license-news-stories