What Is Your Business Data Worth? Start With One Complete Job.

Put one job on the table

Imagine a commercial service company with a recurring equipment fault. Its archive contains the original request, photographs, a technician's observations, parts ordered, a second visit and a later confirmation that the repair held.

Now imagine a second version of the same archive: a spreadsheet of addresses, invoice totals and “closed” statuses.

Both describe the company's activity. They support very different kinds of analysis. The first might help a specialist study diagnostic steps and outcomes. The second could support narrower questions about job timing or billing, assuming the relevant permissions and fields are present.

The point of reviewing a complete job is to discover the capability your records can support. It gives the buyer something more concrete to react to than an impressive count.

Use the “replay test”

Choose an approved example internally and ask a colleague who did not work on it to reconstruct what happened. This is an informal scoping exercise, not a certification or valuation formula.

Write down every place the reviewer has to guess. Those gaps are useful findings. They tell you whether the next step is recovering an export, documenting a field or narrowing the proposed use.

Count useful examples after you define one

Suppose an internal review finds 10,000 closed service cases. Of those, 7,000 have action histories. Only 4,200 also have an outcome that a domain expert can verify.

Those are hypothetical numbers, but they illustrate a better conversation. “We have 4,200 candidate cases with actions and a verifiable outcome” is more informative than “We have 10,000 cases.” It also creates questions: Which cases are missing? Were unsuccessful repairs excluded? Does the coverage favor certain machines or certain years?

If you report a percentage, keep its denominator visible. “Eighty percent linked” is ambiguous. “Eight hundred of the 1,000 sampled eligible work orders linked to a service report” can be examined and reproduced.

Volume still matters when a task requires breadth. The goal is to count the material that fits the task and explain the conditions under which it was counted.

Preserve the things that make the example meaningful

Privacy preparation and usefulness need to be considered together. A replacement identifier can help a reviewer follow one equipment unit across visits. Inconsistent replacements can make that same history look like several unrelated machines. Neither approach, by itself, establishes that the data is anonymous or approved for sharing.

Bring the privacy reviewer and the domain expert into the same conversation. Ask which relationships the proposed task needs, which details can be removed and how the resulting package will be checked. A rare incident or a revealing passage of free text can require attention even when obvious names are gone.

Time deserves the same care. A document written after a repair should not appear to have been available before the diagnosis. If you want to test the next action, exposing the final result would spoil the exercise.

Research on dataset documentation offers a useful foundation: record how the material was created, what it contains and which uses it is suited to. Our practical recommendation is to make that explanation short enough that a buyer actually reads it, with deeper evidence available behind it.

A price belongs to a defined opportunity

An archive may support several possible uses, but buyer demand determines which ones become commercial conversations. A dataset useful for one team may be irrelevant to another team's current project.

Troveo's company-data valuation guide discusses factors including distinctiveness, history, connectivity and outcomes. Ohio Training Data's owner guide likewise starts with task fit and usable evidence. Treat those as questions for investigation, not a price calculator.

Before comparing offers, put them on the same basis. Do they cover the same records, buyer uses, duration and delivery obligations? Is the quoted amount guaranteed, conditional on acceptance or dependent on later sales? Who bears preparation costs? A larger headline can describe a much larger commitment.

The one-page brief worth forwarding

Use these six prompts to prepare for an initial discussion:

1. The job: “Our records may help a developer learn or test ______.”
2. The evidence: Describe one complete example and the systems that contain it.
3. The coverage: State the period, estimated eligible examples and important variations.
4. The limits: Name missing history, inconsistent fields and known sampling biases.
5. The authority: Identify the people responsible for commercial approval, rights and privacy review.
6. The next question: Specify what a buyer or specialist would need to examine next.

That brief is useful even if you decide against licensing. It tells your own team where operational knowledge is well recorded and where it still depends on someone's memory.