Is my company too small to license its data?

Founders of small companies often assume their data is not worth a conversation. Fifteen people, a few years of Slack, one code repository: surely AI labs want bigger archives than that. Sometimes that instinct is right, but often it is not, and the only reliable way to know is to count.

Size matters. It just matters less than most people expect, and it matters alongside several other factors that small teams can be strong on.

Headcount affects value because more people produce more varied work. A 200-person company has more roles, more handoffs and more parallel projects than a 15-person one, and that variety is useful for training agents.

But headcount sits next to three other factors: years of history, the number of systems in use, and connection depth, meaning how often the same piece of work can be traced across several tools. A small company that scores well on those three can end up with a more useful dataset than a larger company that scores poorly on them. We cover how these combine in how company data is valued.

Typical licences range from $20K to $400K+. Where a company lands in that range depends on all four factors together, and there is no firm number before the data is counted.

Small teams have some real advantages. Everyone tends to work in the same few tools, so the links between systems are often tighter than in large organizations, where departments drift onto separate stacks. When five engineers and a product lead share one tracker and one chat workspace, nearly every ticket has a matching conversation.

Small teams also tend to do more of their work in writing. Without layers of management, decisions get made in a channel or a thread rather than in a meeting nobody recorded. That leaves a fuller record of reasoning.

They also cover a lot of ground per person. In a small company, the same few people handle sales calls, support tickets, hiring, vendor negotiations and product decisions. An archive like that shows one team juggling many kinds of work at once, which is close to what general-purpose agents are being asked to do. A larger company might split that work across departments and tools that never touch.

Imagine a hypothetical 14-person software company that ran for five years. It used Slack, Jira, GitHub, Google Docs and a CRM the whole time, and its engineers referenced ticket IDs in every pull request. That company has five systems, five years and strong connection depth. Its headcount is modest, but the dataset could still be well worth evaluating.

Now imagine a 14-person services agency that ran mostly on phone calls and in-person meetings, with email and a shared drive as its only systems. The headcount is the same. The dataset is much thinner, because most of the work never made it into a tool.

Being direct about this saves everyone time. A dataset is less likely to justify a licence when several of these are true:

• The company was active in its tools for only a few months
• Most work happened in person, on calls or in direct messages the company cannot include
• Only one system was used, or the others have already been deleted
• Very few people were active, and the work was mostly one founder's inbox
• Large portions would need to be excluded because of client confidentiality

None of these is automatically disqualifying. A short history across many well-linked systems can still be useful. But if most of the list describes your company, it is worth going in with modest expectations. Our post on the six systems that make a dataset valuable explains which tools tend to carry the most weight.

Small companies have small teams, and nobody has time for a long process. The process is designed around that.

It starts with a 15-minute intro call. If it makes sense to go further, you set the scope and any exclusions, and those are annexed to the contract before any data moves. The evaluation is read-only and takes under an hour of your time. Anonymization happens in-house, you review an anonymized sample, and payment goes to the company entity in two to four weeks. In total, it is about an hour of your time.

That matters for small companies in particular, because the alternative is usually doing nothing. Data sits in a workspace until a subscription lapses, and then it is gone. If the company is winding down, a director, officer, liquidator or assignee with authority over company assets can sign, and the payment still goes to the company entity.

If you are trying to judge your own position before a call, a few questions get most of the way there:

• How many years of history are still accessible in each tool?
• How many distinct systems did the team actually use day to day?
• Do tickets, commits, docs and chat threads reference each other?
• How many people were active in those tools at the company's peak?

Rough answers are fine. Nobody expects you to audit your own workspace before the first call, and the evaluation will measure each of these properly. Answering them does not give you a price, but it does tell you whether the conversation is worth having. The value estimator turns the same questions into a rough range.

If the answers look reasonable, the next step is short. Book a 15-minute call and we can tell you plainly whether your company's data is worth evaluating.