Usually not below about thirty employees, and almost always above a hundred. The product solves a problem that only appears at scale: nobody can see who the company collectively knows. In a fifteen-person company the founders hold that map in their heads and it is accurate. The question is when your head stops being enough.
The threshold is memory, not headcount
The real test is whether one person can still answer "does anyone here know someone at Acme" reliably.
- Under ~30 people. The founders know. Asking in Slack works. A tool would formalise something that is not broken.
- ~30 to 100. It starts failing quietly. The founders still think they know, and increasingly do not, because half the company joined after them.
- Over ~100. Nobody knows, and the relationship that opens an account belongs to someone three teams away who has never been asked.
The dangerous band is the middle one, because the failure is invisible. Nobody reports the introduction that was never made.
Where a small team gets value anyway
Three cases where it earns its place early:
- Enterprise deal sizes with a small team. If twelve named accounts decide your year, finding one path into three of them pays for almost anything.
- A team assembled from strong networks. If your first ten hires came from well-connected companies, their prior relationships are substantial and nobody has mapped them.
- Investor and board networks. Usually offered, rarely mapped, and almost never asked for specifically. A founder who arrives with a name gets an introduction; a founder who asks "can you help with intros" gets sympathy.
What a small team should do instead, first
Before buying anything:
- Ask. In a company of twenty, a Slack message naming ten target accounts works, and costs nothing.
- Map the founders' networks deliberately. Former employers and colleagues, written down once.
- Ask investors for named introductions, not general help.
If those stop scaling - if the Slack message gets ignored, or you cannot remember who you already asked - that is the signal, and it arrives before the headcount does.
The honest answer about cost
Per-seat pricing makes small teams look like a cheap sale. It is also the model that fits this product worst, because value comes from how many people connect their inbox rather than how many have a licence.
A five-seat deployment where only sellers connect produces a thin graph, which produces few paths, which produces the conclusion that the category does not work. If you buy early, connect everyone and buy on a model that lets you.