When not to build custom
Custom software is the most expensive way to solve a problem that already has a product. Sometimes that’s the point — usually it isn’t.
Custom software is the most expensive way to solve a problem that already has a product on the shelf. Sometimes the expense is the point — the process is the advantage, and nobody sells your edge in a subscription. Usually, though, it isn’t. Usually someone just hates the spreadsheet.
It’s worth knowing which situation you’re in before anyone starts building.
The honest test
Three questions. Does an off-the-shelf product do ninety percent of the job? Is the missing ten percent where you’re actually different, or just where you’re used to something else? And would you still want custom software if the product cost nothing?
If a product does the job and the real driver is spreadsheet fatigue, buy the product. Change the process to fit it if you must. That transaction is boring, and boring here means cheap.
Where custom earns its keep
There are exactly a few good reasons. The process itself is the advantage — competitors can’t buy how you quote, plan, or serve, off a shelf. The ten percent gap is the actual work of your business, and no product will ever prioritize your edge. Or the shape is the problem: your operation needs systems that agree with each other, and gluing six subscriptions together with exports and imports and a person retyping in between costs more, month after month, than one thing built for how your company is shaped.
Notice what’s not on the list: preference, habit, and distaste for a vendor’s interface. Those have prices attached too, just hidden ones.
The trap in the middle
The expensive place is the middle — customizing around a product you never liked until, two years later, you’ve built software anyway. Badly, at integration prices, spread across invoices nobody connected.
The rule of thumb is blunt. If you’re three spreadsheets and two workarounds deep into making a product fit your company, either change the company’s process to match the product, or admit you need the custom thing and build it properly. Halfway is where money goes to die quietly.
The cheapest software is the kind you don’t build. The second cheapest is the kind you build only where it matters — and buy the rest like everyone else.
Working on something this applies to? Tell us about it — thirty minutes, no pitch.