Almost every services firm eventually tries to build a product. The logic is sound: services revenue is linear in headcount, product revenue is not, and the team has already built the same thing four times for four clients. The attempt usually fails, and it fails in a consistent way that is worth understanding before starting rather than after.
The mechanism is straightforward. Client work has a deadline, an invoice and a person asking about it. Product work has none of those. When a client escalation lands, the rational short-term decision is to move an engineer off the product, and that decision is correct every single time it is made. Repeated for eight months it produces a product that is perpetually two months from launch. Nobody chose that outcome; it emerged from a series of individually sensible choices.
The structural fix is separation strong enough to survive pressure. A dedicated team that is not on the services rota, with its own targets and a leader with authority to refuse client work. Ring-fenced budget treated as spent rather than as capacity available for reallocation. And an explicit decision at the top about what happens when the two conflict, made in advance and in writing, because made in the moment it always resolves toward the client.
The second failure is subtler and concerns what actually gets built. The productised version of four bespoke projects is usually a configurable system that does all four, which means it is complex, hard to sell and hard to demonstrate. Products succeed by doing one thing for many customers, not many things for a few. The discipline is to look at the four projects and identify the narrow common core, then build only that and refuse the configurability that made the services work valuable.
The commercial muscles are different too, and this is underestimated. Services sell through relationships, proposals and trust built over meetings. Products sell through positioning, self-serve trials, documentation, pricing pages and marketing — competencies a services firm generally does not have and does not develop by assigning the task to whoever is free. Budgeting for product marketing as a real function, not an activity, is part of the cost of the attempt.
None of this argues against trying. It argues for entering with clear eyes: a separated team, a narrow first product, a realistic runway measured in years rather than quarters, and an honest internal agreement about what the firm will sacrifice to protect it. Firms that make that agreement explicitly have a reasonable chance. Firms that treat the product as something the team will get to between projects are, in practice, deciding not to build it.
