How do I know if this enterprise deal is a real repeatable market or a one-off custom project that will trap me?
The short answer
If closing the deal requires you to build something only this customer needs, you're taking on a services project wearing a product costume, and it will eat the roadmap. Test repeatability by asking whether three other companies would buy the same thing without changes, and whether you can charge for it as a product rather than for your time. Take the occasional custom deal early for cash and learning, but cap it, keep the IP, and refuse to let one big customer's wishlist become your entire product plan.
Go deeper, your way
3 hand-picked resources, 2 link-checked. Pick how you want to dig in.
📄 Article
✓ Link checkedFreeIntermediate
Why we picked it
When half your revenue comes from one customer's custom asks, the real risk is not the roadmap slipping, it is quietly becoming a dev shop wearing a software costume. This piece names that trap plainly and gives you a litmus test: if you added 100 customers tomorrow, would software handle them or would people? That is the question to hold up against every custom feature that big customer requests.
Why we picked it
When the big customer's wishlist lands on your desk, this gives you a hard gate instead of a gut call: build the custom feature only if it is already on your 24-month roadmap, the contract is 1+ year and paid a lot (never for a pilot), the customer does not demand exclusivity, and you have met them yourself. Lemkin's reframe is the repeatability test in action: if one enterprise wants it and it fits your roadmap, at least nine more will, so it stops being custom and becomes product direction. His "do 1 to 2, not the 10 to 12 sales asks for" line is the discipline that keeps one account from eating the plan.
Only build the custom ask if it is already on your roadmap AND a paid 1+ year contract, otherwise it is a services project in disguise.
If a big customer wants something on your roadmap, treat it as a paid preview of what nine other big customers will want, which is the definition of repeatable.
Sales will surface 10 to 12 custom features a year; you can only afford 1 to 2 that reshape timing, so the CEO, not sales, holds the gate.
Why we picked it
Before you say yes to the custom features nobody else wants, this essay makes you sit with the real risk: one account that pays a lot can quietly take over your roadmap and turn you into its in-house vendor. McAfee tells the story of a company that let one customer reach 80 percent of revenue, bent the product to that customer's demands, and collapsed when the customer's funding dried up. It is a starting point for weighing when a big customer's asks are worth it and when they are a trap dressed as a win.
A single dominant customer can mask the fact that you do not yet have real product-market fit, so the revenue feels like validation when it is actually dependence.
Every custom request pulls capacity away from the product you meant to build; if you lose that one customer, you lose the revenue and the roadmap you traded away for it.
Keep some engineering capacity reserved for work that serves the whole market, not just your biggest account, and be wary of open-ended custom scopes.
