Scope for the business you have, not the one on your roadmap

There's a pattern we see in nearly every discovery call: someone describes a workflow that breaks every Tuesday, then asks us to build a system that handles ten times the volume, three new business lines, and a partner integration that may or may not happen next year.

We get it. If you're going to pay for custom software, you want it to last. But over-scoping for a hypothetical future is one of the most expensive mistakes we see, and it's gotten worse now that AI-assisted development has made custom builds accessible to smaller teams.

The build vs. buy math actually changed

CIO.com recently argued that leaders should reopen the build vs. buy question because AI-assisted development has meaningfully shifted the economics. That's true in our experience. Things that used to take a quarter take a couple of weeks. Things that used to require a dedicated engineer can be maintained by a small team on managed infrastructure.

But here's the second-order effect nobody talks about: when building gets cheaper, the temptation to build *more* gets bigger. People who would have bought a SaaS subscription and lived with its limits now ask for a custom system. And then they ask for that system to do everything the SaaS did, plus everything they wished it did, plus everything they might want in 2027.

That's not a custom build anymore. That's a platform. And most businesses don't need a platform.

What 'fit' actually means

In a separate piece, a CIO walked through the practical decision maze of build vs. buy and landed on a useful frame: fit isn't about features, it's about how much of your business logic the tool can hold without distortion.

Same principle applies to custom builds. A well-scoped custom workflow fits your business the way a tailored jacket fits a person. A badly scoped one fits the business you described to a consultant in a workshop two years ago.

The question we ask on every discovery call is some version of: what does this look like on a normal Tuesday? Not the launch. Not the IPO. Not the acquisition. Tuesday.

The boring scope is usually the right scope

Here's what we've found works:

- Build the thing that handles your current volume with maybe 2x headroom. Not 10x. - Encode the rules you actually follow today, not the ones you aspire to. - Leave the integrations you're not sure about as manual steps. They're cheap to automate later. They're expensive to build wrong now. - Skip the admin panel for the edge case that happens twice a year. Do it by hand.

This sounds like a smaller engagement, and it is. That's the point. The version of your business that exists in 18 months will know things you don't know today. Building for that business now means guessing, and the guesses are usually wrong.

The best custom software we've shipped is the kind clients forget is there. It handles the boring stuff. It fits. When the business changes, we change it. That's the whole loop.