Seller intent
How AI data valuation actually works
Last updated /Seller intent
AI data valuation, on this site, means a structured reading of what a company has and whether that picture could matter to a buyer. It is not a certified appraisal. It is not a broker's opinion of a sale price. It is not an offer. The output of the first step is an estimate you can ignore.
People often arrive wanting the dollar answer. The method matters more, because the method is what keeps the dollar from being a guess. If you want the factors in plain language, read how much your company data could be worth. This page is the procedure: what gets written down, in what order, and what is deliberately left out.
A valuation that skipped the permission question would be unfinished. A license is a defined permission, and a number that assumes you can grant that permission is not a valuation. It is a wish.
Why the method is a profile, not a file weigh-in
Weighing files rewards hoarding. A profile asks whether the history is usable. The fields are ordinary on purpose: company, industry, employee count, operating history, digital history, estimated volume, systems, categories, language, geography, rights status, sensitivity, and whether a later export looks possible. Those are the same kinds of fields a Data Partnership Profile carries. None of them is a price.
Two companies can share an industry and a headcount and still produce different profiles. One has a long digital history in a few systems of record. The other has a short export, three overlapping tools, and a contract that forbids reuse. The method has to be able to say those are different cases. A single industry average cannot.
What the review is allowed to treat as evidence
Evidence is something you can point at. "We have a lot of data" is not evidence. "We have twelve years of closed work orders in one system, and the closed reason is filled in" is evidence. The eight steps from discovery to a possible close start with discovery and a count, before anyone packages a profile.
- Coverage. Which years, which teams, which systems. Gaps are part of the answer.
- Categories. Communications, operations, customer interactions, technical records, documents. A mix is described, not praised in the abstract.
- Outcomes present or absent. The profile should say so, not imply a happy ending.
- Sensitivity. What would have to be removed or left behind.
- Rights status. A plain statement of what is known and what still needs counsel. Not a legal conclusion from us.
- Readiness. Whether a defined slice could be prepared later, if you say yes.
What we refuse to treat as evidence
A competitor's rumor. A viral post about what "AI companies pay." A row in the illustrative industry benchmarks. Those benchmarks are labeled as benchmarks so nobody confuses them with a valuation of a named company. Your company's data could be worth up to $5M+ only as the public ceiling. It is not an input the method pastes into your result.
What makes a valuation thin or misleading
Missing history, unnamed systems, and a rights answer of "we think it's fine." Double-counting the same records because they were copied into a second tool. Treating personal data as a bonus instead of a constraint. Quoting a timeline. Comparing your estimate to someone else's offer, which we will not do, because we do not publish other companies' deals.
A thin valuation is also what you get if the only input is the industry. We would rather hand back a wide, humble range, or a "not a fit," than a precise number built on a category.
The two layers: the check, then the profile
- Free check. You answer questions. No files move. You see an estimate and the disclosure that estimates are not offers.
- Qualification. We ask whether the business matches the demand we can actually describe. Size, history, and the kind of work matter. About 50 or more employees is the usual fit, not a magic threshold that creates value.
- Profile. If you continue, the answers become a Data Partnership Profile a buyer could read without the raw files.
- Anything after that, including a conversation about terms, is optional and visible to you before it is final.
Illustrative example, not a customer: two profiles both say "manufacturing" and "about 80 employees." One lists a single maintenance system with a decade of closed orders and a note that customer names can be dropped. The other lists six tools, two years of partial exports, and an unknown position on vendor contracts. The method is doing its job if those profiles do not receive the same sentence.
Questions people ask
Is this a formal appraisal?
No. There is no certified appraisal, no stamp, and no number you can take to a lender. It is an estimate from a questionnaire, and later a profile if you want one.
Do you use the industry table as the valuation?
No. The Business Data Value Index is illustrative. A valuation uses your history, systems, categories, and rights notes. The table is not an input we copy.
Can the estimate change after the profile?
Yes. The check is a first reading. A profile can show that the history is deeper, or thinner, than the first answers suggested. Neither version is an offer.
Who sees the valuation?
You see the estimate. A buyer does not see your files at that stage. If a profile is shared later, it is the profile you approved, not a surprise attachment.
Will a higher score guarantee a buyer?
No. A cleaner profile is easier to discuss. Demand still has to exist, and you still have to agree. We do not sell certainty.
What should I do with a low result?
Read it. A low result or a poor fit is allowed to be the end. You do not owe a follow-up call.
See what your data could be worth
Your company's data could be worth up to $5M+.
Potential licensing value varies by dataset, rights, quality, volume, buyer demand, and transaction terms. Estimates and examples are not guaranteed offers.
Check my data value