A data license should begin with a written scope, not a data export. Your team should be able to identify the source systems, record types, date ranges, excluded fields and review steps before any transfer takes place.
Confidentiality and security
What to settle before you share anything
Define the dataset precisely
Name the systems and record types involved. Separate useful operational patterns from customer identities, credentials, financial details, health information, pricing and anything your company cannot share.
Remove more than obvious identifiers
Names and email addresses are only the beginning. Free text can contain account numbers, private links, signatures, employee names or customer-specific details. Review both structured fields and unstructured text.
Agree on review and approval points
Document who performs de-identification, who checks the result and what happens if sensitive material remains. If your team wants to approve a sample or the final dataset, put that step into the agreed process.
Limit the transfer itself
The agreement should cover delivery method, access controls, retention and deletion. A carefully scoped dataset still needs a carefully controlled handoff.
A practical review checklist
Four things to take into the room
- List included systems, fields and date ranges.
- Create an explicit do-not-share list.
- Test de-identification on unstructured text as well as database fields.
- Write down who reviews, who approves and how exceptions are handled.
The Datumi process
How Datumi approaches confidentiality
Datumi starts with an inventory: approximate volumes, systems and date ranges. The qualification process does not ask for records or exports.
Preparation and de-identification happen within an agreed scope before buyer review. The license then defines the dataset and permitted use, rather than granting open-ended access to your systems.
“Can your team describe exactly what would leave—and point to the step where it is checked?”
Explore your company’s data potential
Start with a free data value check. It asks about your systems and approximate record volumes—no files or data exports. It is an initial assessment, not an offer or a guarantee.
Next question · 02
Who gets to use your data—and for what?