The Temptation Sounds Reasonable
This comes up in B2B constantly. A customer wants something specific. The request is not crazy. In fact, it often sounds very reasonable. Add this integration. Support this workflow. Expose this data. Match this internal process. Everyone around the table can explain why it matters.
That is exactly why it is dangerous. The most painful mistakes in product are rarely obvious mistakes. They are the ones that sound sensible enough to keep getting approved.
You Are Not Supposed to Make Everyone Happy
A lot of teams still behave as if good product management means satisfying every serious request from the market. That is not normal. You are never going to build a product that makes everybody happy. If some customers are frustrated, that does not automatically mean you are failing. Usually it means the market is heterogeneous and your product has boundaries.
That is healthy. In B2B, the mistake is treating every unhappy edge case as a roadmap obligation. If you do that long enough, you do not get a sharper product. You get a pile of accommodations.
Custom Work Often Borrows Someone Else's Value
This is the part people miss. A lot of so-called custom product work is not actually creating a new center of value. It is extending a process or product where the customer already had the aha moment somewhere else.
They already understand why the core system matters. They already use it. They already depend on it. Your work is being pulled in to make that existing system fit a little better into one more workflow, one more team, one more internal process. That can absolutely bring in revenue. It can also quietly trap you.
Why It Feels Attractive
From the inside, this kind of work feels safer than it is. There is a customer attached to it. There is urgency. There is a champion. Sometimes there is immediate money. Compared to ambiguous product bets, custom work feels concrete and rational.
But the problem is not whether you can justify each request one by one. The problem is what happens in aggregate. Each request pulls the product slightly toward a workflow you do not own, slightly toward a process you did not design, slightly toward another company's source of value. Then you wake up one day and realize you have spent two years building a better attachment to somebody else's product.
How Companies Lose Years
Very few teams decide up front to work on the wrong thing. What happens instead is slower. A company keeps saying yes to adjacent work because each individual step looks defensible. A customization here. A special integration there. A workflow adjustment for an important account. Nothing looks fatal in isolation.
But the direction is wrong. Thinking a little harder at the beginning would save a lot of teams years of pain here. The question is not "can we sell this?" The question is "does this move us toward our own product, or deeper into supporting someone else's?" Those are very different futures.
Money Makes This Harder
This is also why revenue can be misleading. Custom work often arrives with money attached, which makes it feel validated. But revenue alone does not tell you whether you are strengthening your product or just renting out your team to improve a workflow that fundamentally belongs somewhere else.
That distinction matters a lot. The market will happily pay you to drift.
A Better Filter
I think there is a useful filter for this category of decision. If the work mostly expands another product's usefulness, another team's process, or another system's value moment, you should be suspicious. Not because it is useless. Because it is probably not your product.
The Line
Custom work is not automatically bad. Sometimes it teaches you where the real product is. Sometimes it opens a wedge.
But most of the time, especially in B2B, it is much easier than people admit to confuse customer proximity with product clarity. Any team can spend years building respectable, revenue-generating, deeply justifiable things that still turn out to be the wrong project.
That is why product boundaries matter, and that is why "the customer asked for it" is nowhere near a good enough reason.