Operating companies are joining the data supply

Data licensing is usually discussed as something that happens at the end of a company's life, when the tools are about to be cancelled and the archive is about to disappear. That framing leaves out a large part of the picture. Companies that are very much still operating hold the same kinds of records, and they can license a carefully chosen slice of them too.

It is easy to see why closing companies were the obvious starting point. Their data has nowhere else to go. Subscriptions lapse, workspaces get deleted, and years of operational history vanish unless someone acts. For a board or liquidator winding things down, licensing turns a disappearing asset into a recovery for the company.

But the underlying value has nothing to do with whether a company is closing. AI labs want operational data because it shows how real work gets done across chat, email, docs, tickets, code, CRM, support and finance tools. An operating company's archive shows exactly the same thing, often with more years of history and a tool stack that is still in wide use.

The difference is not in the data. It is in how much of it the company is comfortable sharing, and on what terms.

An operating company does not need to license everything, and usually should not. The more natural approach is to choose a slice that is valuable to a buyer and low in sensitivity for the business. Common shapes include:

• A historical date range, such as everything older than a chosen cutoff
• A retired product line or a project that has been completed
• A single team or function, such as engineering tickets and code, without sales or finance
• A business unit or office that has since been closed or restructured
• Tools the company has migrated away from, where the old workspace still exists

Each of these is set as scope before any data moves, and the scope is annexed to the contract. The defaults still apply on top: HR, payroll, medical and legally privileged folders are out, and direct messages stay out unless the company chooses to include them and has the right to.

Imagine a 90-person software company that moved from one ticketing tool to another two years ago. The old workspace holds five years of tickets linked to code and chat. Licensing that archive, with a cutoff before the migration, touches nothing the team uses today.

The process is the same as for any seller: a 15-minute intro call, scope and exclusions, a read-only evaluation that takes under an hour of your time, in-house anonymization, a review of an anonymized sample, and payment to the company entity in two to four weeks. But a few considerations carry more weight.

Current customers matter more. A closing company's client relationships are winding down; an operating company's are live. Reviewing confidentiality clauses and excluding particular clients' channels and threads is often the first scoping step. Customer confidentiality clauses and your data goes into this.

Current employees matter more. Many of the people who wrote the messages in the archive may still work there. Anonymization replaces every person with a consistent token, and the licence forbids anyone from marketing to, profiling, contacting or re-identifying them, but telling the team plainly is still the right thing to do.

Competitive sensitivity matters more. Plans and pricing discussions from last quarter are more sensitive than those from five years ago. A historical cutoff handles much of this on its own, and specific channels or folders can be excluded on top of it.

From the lab side, operating companies help balance the supply. A training mix drawn only from companies that closed would lean toward businesses that did not make it, and toward whatever stages and sectors those closures happened to come from. Adding companies that kept working gives a fuller picture of how work gets done.

Operating companies also tend to offer longer histories. A company that has run for a decade, even with a cutoff a couple of years back, may hold more years of connected records than a younger company that is shutting down. Years of history and connection depth are two of the main drivers of value, alongside headcount and the number of systems.

And operating companies are more likely to run tool stacks that are still in common use, which matters to labs building agents for today's workplaces. For more on how buyers use this material, see who buys company data, and what they do with it.

For an operating company, licensing a slice of historical data is a commercial decision like any other. The questions are whether the value justifies the internal attention, whether the scope can be drawn tightly enough to keep the business comfortable, and whether the board and team are on board. Five questions your board will ask about licensing data is a useful place to start that conversation.

Typical licences range from about $20K to $400K and up, and no firm number is possible before the data is counted. Because the evaluation is read-only and takes under an hour, finding out costs very little, and nothing moves until the scope and contract are agreed.

If your company is operating and you want to know what a scoped licence could look like, book a 15-minute intro call.