Essay · October 27, 2025
Two to five people
Every few months someone asks why we don't grow. The honest answer is arithmetic. At five people, everyone on an engagement has read the source data: the actual parcels, the actual work orders, the scanned as-built with the coffee ring on it. At six, someone is managing instead. At ten, the person who talks to the client and the person who reads the data have never met, and the firm has started manufacturing the exact problem clients hire us to escape.
What the buyer is actually buying
A director who hires a boutique isn't buying capacity; they can get capacity anywhere, in bulk, with a rate card. They're buying the collapse of distance between the conversation and the work. When they describe a problem on a call, the person nodding is the person who will be inside the database that afternoon. Every layer of delegation restores the distance and dilutes the thing being purchased.
What breaks at six
Ownership, mostly. At five, "whose fault is this" has an answer that is also "who will fix it." Documentation is written by the person who will be embarrassed if it's wrong. Handoffs happen inside one working memory. We've watched excellent eight-person teams spend a third of their hours reconstructing context that a four-person team never sheds.
The confession
Small has costs and we should name them. We say no a lot, to good projects, at painful moments. Two engagements a quarter is the ceiling, and a single illness reshuffles everything. Once, we took a third engagement because the client was persuasive and the work was interesting. Quality held; the people didn't. Nobody slept well for six weeks, and the retro afterward wrote the rule we now keep: the cap is the cap.
We are not against big firms. Some problems genuinely need fifty people. But the problems we love (specific, constrained, on the ground) need five or fewer, fully present. So that's the size of the practice, on purpose, indefinitely.