Compare the same product
Two offers are not comparable if one covers a narrow evaluation license and the other includes perpetual commercial training rights, affiliates, and updates. Write a scope sheet for each proposal: collection version, eligible volume, use, term, territory, exclusivity, redistribution, and services.
Ask which unit controls payment. A file, token, hour, resolved thread, accepted example, and dataset license measure different things. If the buyer pays only for accepted units, define acceptance and estimate the eligible volume rather than multiplying a headline rate by every item in storage.
Build a net-proceeds view
Estimate preparation, legal review, annotation, security work, delivery, support, and tax-administration costs. Keep uncertain costs as ranges and state their assumptions. Do not treat optional future orders as guaranteed revenue. An offer with a larger headline amount can produce less value if it requires extensive uncompensated work.
Consider timing and risk. A payment contingent on undefined buyer satisfaction differs from a milestone payment after objective checks. Separate guaranteed consideration from usage-based amounts, renewals, and bonuses. Ask how each contingent amount will be measured and verified.
Value the rights you give up
Exclusivity can restrict later opportunities; broad sublicensing can expand the audience beyond the team you evaluated. A perpetual right to retain trained models may have consequences beyond a short raw-data access period. Write these terms plainly before deciding whether the price compensates for them.
Use alternative offers only when they are real and comparable. Do not infer a universal rate from a public announcement involving a major publisher or a different data category. Public deal values may include product access, distribution, services, or other terms that are not visible.
- Compare paid units after exclusions.
- Separate one-time revenue from contingent payments.
- Document seller costs and continuing obligations.
Use negotiation to reduce uncertainty
Ask for a paid pilot, a narrower initial license, clearer acceptance criteria, or separately priced updates when these address the main uncertainty. A smaller well-defined deal may be preferable to an ambiguous promise of scale. Decide in advance which uses or risks are outside your acceptable scope.
This is a commercial comparison framework, not a valuation or tax opinion. Use relevant professional advice for the actual agreement. A defensible decision states why an offer fits the collection, rights, costs, and alternatives; it does not depend on an invented industry average.
Common questions
What is my company’s data worth?
There is no reliable universal answer from storage size alone. Task fit, usable volume, rights, quality, buyer needs, and contract scope all affect an actual proposal.
Can I use a public licensing deal as a price benchmark?
Only with caution and explicit limitations. If scope, services, exclusivity, and consideration are undisclosed, the comparison may not be meaningful.