Questions to ask before licensing company data

What data are we actually talking about?

Think about a support ticket: a customer asks a question, your team investigates and someone records the fix. That is a recorded example of how work gets done.

Business records can include support conversations, project discussions, sales activity, process documents and approved call transcripts. This is not an invitation to hand over your entire inbox or a list of people to contact. Start with specific records and a clear proposed use.

Do I have to send my files to get started?

With Maryah, the initial assessment starts with counts and screenshots, not a raw dataset. Data is transferred only after a buyer deal is agreed.

Screenshots still need care. Hide names, email addresses, customer details and other sensitive information before sharing them. An assessment should not become an accidental disclosure.

Does owning the business mean I can license everything?

Do not assume that. Your records can contain information belonging to customers, employees, suppliers or other parties.

Ask your legal adviser to review the relevant agreements, privacy obligations and rights before anything is licensed. The question is not just “Can we export this?” It is “Are we allowed to use and share this information for this purpose?”

Is replacing names enough to make data anonymous?

No. Replacing “Jim Smith” with “Person 14” illustrates one step, not the whole privacy process.

A distinctive job title, event or combination of details may still identify someone. Ask how identification risk is assessed, who reviews the output and what evidence supports the result. Be cautious about any promise of zero risk.

What if the records reveal our competitive advantage?

Protecting people’s identities and protecting business secrets are different questions.

A record might hide every name and still reveal a pricing method, a negotiation strategy or a valuable technical process. Identify what you would never want disclosed before discussing a dataset. Some records may be unsuitable for licensing.

What will a buyer do with it?

AI uses can include training a model, improving an existing model or making synthetic practice examples. The exact permitted uses should be defined in the agreement.

Ask who receives the data, whether it can be shared onwards and whether restrictions cover uses that concern you. Do not treat “for AI” as a complete explanation.

Could this help AI compete with us or change jobs?

It is a fair concern, not something a payout makes irrelevant. Consider which capabilities the records could help develop and whether that is acceptable to your company and team.

No general marketing promise can guarantee that future AI systems will never compete with your work. If the proposed use conflicts with your business interests or values, declining is an option.

Why is the offer so high? Is it a real offer?

Separate a calculator estimate from a buyer’s written offer. An estimate is an indication, not a commitment to pay.

Before comparing amounts, check the dataset, permitted uses, duration, exclusivity and payment conditions. A larger figure does not necessarily mean better terms. Ask what has actually been evaluated and what could change the amount.

Could a one-off payment limit future opportunities?

Check whether the rights are exclusive, how long they last and whether the buyer can grant rights to others. Ask what you can still do with the same records afterwards.

If a company sale is possible, involve the people advising on that transaction. Existing licences should not be an unwelcome surprise during diligence.

What happens if personal information slips through?

Ask who handles the incident, who pays for claims and which obligations survive the agreement. Have your legal adviser examine warranties, indemnities, liability limits and responsibility for each stage of processing.

A statement that data will be cleaned does not answer those contract questions. Neither does a large payment.

What happens when the licence ends?

Ask separately about the dataset and anything created using it. What must be deleted? What can be retained? Can future licensing stop? What rights survive?

Deleting a file is not the same as undoing training already performed. Do not assume that ending a licence automatically removes trained models or derivatives. The agreement needs to address these issues before you sign.

Our software providers already hold this data. Is this different?

Using a tool to run your business and licensing records for another purpose are separate decisions. One should not be treated as permission for the other.

Review the proposed purpose on its own merits. “Someone already has it” is not a reason to skip questions about rights, trust or risk.

I am an adviser. Can I introduce a client without sharing their files?

An introduction and a data transfer are different things. Ask the client whether they want to explore the opportunity. Do not forward client inboxes, deal-room documents or confidential records as part of an introduction.

For M&A advisers, accountants and investment firms, the first conversation can be about the client’s business, the tools they use and whether they want an assessment. Get permission before sharing client information.

Where should we start?

Begin with questions, not a submission. Ask Maryah about an initial assessment using counts and carefully redacted screenshots. Then evaluate the proposed dataset and terms with the people responsible for your company’s legal, security and commercial decisions.

You do not have to share your records to decide whether the conversation is worth having.

This guide provides general information, not legal advice. The right decision depends on your records, agreements and applicable law.