Getting started
How to value company data: cost, market and income approaches
By SourceX Editorial · Reviewed by Noah Loul ·
Short answer
Data valuation methods follow the three classic approaches: cost (what records took to create or would take to recreate), market (what comparable data has fetched) and income (the cash the data is expected to produce). For licensing to AI developers, treat all three as planning ranges. An actual price is set only when a buyer engages with a scoped package.
Key takeaways
- The cost approach measures the effort to create or recreate records, not what a buyer will pay for them.
- The market approach is weak for operating records because comparable deal terms are rarely public.
- The income approach depends on license forecasts that cannot be tested before a buyer shows interest.
- Drivers such as uniqueness, rights and recency explain why two similar archives can be worth very different amounts.
- Report any internal estimate as a range or a floor, and let buyer engagement set the licensing price.
Why is company data hard to value?#
Company data is hard to value because the same records can be worth very different amounts depending on who uses them, for what purpose and under which rights. Years of support tickets may mean nothing to the financial statements yet interest an AI developer building a support agent for the same industry.
Data also lacks the anchors other assets have. Most operational records have no active public market, deal terms are usually confidential, and the cost of creating the records was spent on running the business rather than on producing data.
In most cases, internally generated data may not appear as an asset on the balance sheet, so the finance team often starts with no book value at all. Ask your auditor how any related costs are treated before you report a figure to anyone.
How do the cost, market and income approaches compare?#
The cost, market and income approaches answer different questions: what the data took to make, what similar data has sold for, and what cash it will produce. Appraisers often use more than one and reconcile the results.
For a licensing decision, the approaches are most useful for setting expectations and floors. None of them predicts the number a buyer will sign, because that number depends on the scope, exclusivity, term and use the buyer needs.
| Approach | When it is used | Main inputs | Limits for licensing data |
|---|---|---|---|
| Cost | Internal business cases, insurance, a negotiating floor | Staff time that produced records, system and storage spend, cost to recreate | Measures effort, not demand; old records may have cost a lot and still interest no one |
| Market | When comparable transactions or price lists exist | Prices or terms of similar data deals, adjusted for differences | Few comparables; most data deal terms stay private |
| Income | Acquisitions, impairment testing, licensing forecasts | Expected revenue or savings, useful life, a discount rate | Needs forecasts that depend on buyer interest you cannot yet see |
How does the cost approach work for business records?#
The cost approach values business records by what it took to create them or what it would take to recreate them today. For a services company, that means the hours experts spent writing job notes, RFI responses, support replies or quality reports, plus the systems that held them.
Recreation cost is often the more telling version. A model developer cannot pay people to relive years of real dispatch decisions or warranty disputes, so historical records with real outcomes are hard to reproduce at any price. That gap is one reason the cost approach understates some archives and overstates others.
- List the record families and the systems that hold them.
- Estimate which staff roles produced each family and roughly how much of their time it took.
- Add system, storage and administration spend attributable to the records.
- Note which records could be recreated synthetically and which could not.
- Record the result as a floor and label it as a cost estimate, not a price.
Why is the market approach thin for operating data?#
The market approach is thin for operating data because few transactions are public and those that are rarely disclose terms. Published deals tend to involve media archives or public content, which differ from internal support, engineering or job records in rights, privacy work and intended use.
Adjusting a comparable means correcting for differences in scope, exclusivity, term, record type and preparation. With operating records, the adjustments are usually larger than the comparable itself, so the result says little. Treat any figure from a vendor pitch or press story as anecdote, not evidence.
What does the income approach need?#
The income approach estimates the present value of cash the data is expected to generate through licenses, products or savings. Practitioners use variants such as relief from royalty, which asks what the owner would pay to license the asset if it did not own it, and with-and-without analysis, which compares results with and without the data.
Every income method needs a forecast. Before any buyer has looked at a scoped package, that forecast rests on assumptions about demand, term and exclusivity the company cannot test. For internal uses, such as faster estimating or shorter support queues, the income approach is more workable because the savings show up in the company's own numbers.
Which factors move licensing value?#
Licensing value moves with the qualities of the records, not with what they cost to produce. The SourceX Enterprise Data Value Framework, a SourceX-developed methodology rather than an industry standard, names the drivers and rates them qualitatively, without prices or index values.
A CFO can use the drivers as a checklist alongside any of the three approaches. They explain why an archive with a high cost basis can still attract little interest, and why a modest one can stand out.
| Driver | Effect | Question for the finance team |
|---|---|---|
| Uniqueness | Increases value | Could a buyer find similar records elsewhere? |
| Domain expertise | Increases value | Were the records written by skilled staff doing expert work? |
| Human-generated signal | Increases value | Did people write the records, or did systems generate them? |
| Scale | Increases value | How many records and years can still be exported? |
| Recency | Increases value | How current is the latest history? |
| Data cleanliness | Increases value | Are fields consistent and records linked to each other? |
| Rights | Increases value | Does the company clearly own and control the content? |
| AI utility | Increases value | Do records show a request, a decision and an outcome? |
| Exclusivity | Increases price | Is the company willing to grant exclusive use? |
| Reproducibility | Reduces value | Could a buyer generate similar data synthetically? |
| Preparation cost and privacy burden | Reduce net value | How much work removes personal and confidential details? |
Illustrative: a 3PL's CFO answers the board#
Illustrative: the board of a fictional regional 3PL asks its CFO what the company's warehouse data is worth after a vendor pitch mentions AI. The archive includes WMS order and exception records, carrier claims, customer service email and dock scheduling history.
The CFO builds a cost estimate from staff time and system spend and labels it a floor. A market search finds no comparable operating-data deals with disclosed terms. An income model would need license assumptions nobody can support, so the CFO declines to build one.
The board receives a cost floor, a qualitative read on each value driver, and a recommendation to run a metadata-only fit check. It gets a defensible answer without a number that could later look invented.
How SourceX approaches data value#
SourceX does not publish prices, payouts or index values. It rates records against the SourceX Enterprise Data Value Framework, then tests interest with buyers using a scoped description of the package rather than the records themselves. Value is known only once a buyer engages, and the supplier approves every step before anything is shared.
Frequently asked questions
Can we record our data as an asset on the balance sheet?
Usually not just because it might be licensed. Accounting rules generally limit recognition of internally generated intangibles, while data purchased or acquired in a business combination can be treated differently. Ask your auditor and tax advisor how your situation is handled before you report any data value externally.
Should we hire an appraiser to value our data?
An appraiser helps when a value must stand up to outside scrutiny, such as a purchase price allocation, a tax filing, a lender request or a dispute. For a first licensing decision, a cost floor plus a qualitative driver review is usually enough, because buyer engagement sets the actual price anyway.
Does more data always mean more value?
No. Scale is one driver among several. A smaller archive of expert-written records with clear outcomes and clean rights can be worth more than a larger set of system logs, duplicate email or records that need heavy privacy work before anyone can use them.
How does exclusivity change the value?
Exclusivity tends to raise the price a buyer will pay, because it keeps competitors away from the same records. It also limits the company's future options, including licensing the same scope to others. Consider limiting exclusivity by field of use and term rather than granting it broadly.
Why not use our storage costs as the value?
Storage costs measure what it takes to keep records, not what they are worth to anyone. Cheap storage can hold valuable records, and expensive systems can hold records nobody wants. Storage spend belongs in the cost approach as one input, not as the answer.
Related resources
See if your company qualifies
A short company assessment. No data uploads are needed.