A confession from an agency that builds software for money: about half the businesses who come to us asking for custom software should not buy custom software. Not yet, and sometimes not ever.
We could quietly take those projects. Instead we've turned the diagnosis into an article, because the businesses we talk out of building things have a habit of coming back later, with the right project, and with the kind of trust that no portfolio buys.
So here's the honest diagnostic, the same one we run in discovery calls.
The pattern: chaos arrives dressed as a software request
It usually sounds like this: "We need an app to manage our orders, everything is scattered, staff forget things, customers fall through the cracks."
Notice what's actually being described. Not a missing tool, a missing process. Orders live in WhatsApp, in a notebook, in two people's heads. There's no agreed way anything flows; there's just heroic memory, and heroic memory has started dropping things.
Here's the uncomfortable mechanism: software doesn't create order, it enforces order that already exists. Point custom software at chaos and you get faster, more expensive chaos with a login screen. The app can't decide who confirms an order, when a job counts as done, or what happens when a customer changes their mind. Those are decisions, and decisions are free. They're also the actual product most people are trying to buy when they say "app."
The spreadsheet test (run this before spending anything)
Before commissioning software, prove the process in the cheapest possible material. Take your scattered workflow and force it into one shared Google Sheets file or an Airtable base: one row per order, columns for status, owner, dates, amounts. Then run the business on it, with actual discipline, for one month.
One of two things happens, and both are victories:
It works. The chaos was never a software gap, it was an agreement gap, and a free tool plus a defined process just solved a problem you were about to spend serious money on. Keep the money. Grow. The spreadsheet will tell you loudly when it's dying.
It works, but it strains. The rows hit thousands. Three staff overwrite each other. You're copying the same data between four sheets, customers keep asking for updates you have to type manually, and someone made a decision on last week's numbers because nobody refreshed the file. Congratulations, you now have something priceless: a proven process with documented breaking points. That's not a failed spreadsheet. That's a free requirements document.
When custom software is genuinely the answer
The real signals, as opposed to the mood:
1. The spreadsheet is straining in specific, nameable ways. Not "it feels unprofessional", concrete failures: concurrent edits colliding, manual copying between systems eating hours, no way for customers to see their own status. Specific pain converts directly into specific features; vague pain converts into expensive guesses.
2. Your process is stable and repeats. Software freezes a workflow in code. If you're still changing how you operate every month, freezing is premature, you'll pay to rebuild the freeze. Build when the process has stopped surprising you.
3. The workflow itself is your edge. If how you operate is why customers choose you, your turnaround speed, your quality checks, your delivery logistics, then generic tools force you to work like everyone else, and custom software is how your advantage scales instead of diluting. (Whether to build that or buy something close is the build vs. buy question, which deserves its own article and got one.)
4. The manual hours have a price tag that beats the build cost. Count honestly: hours per week spent on copying, chasing, reconciling, multiplied by what those hours cost, multiplied by 52. When that number embarrasses the software quote, the decision has made itself.
The sequence that never fails
If you remember one thing: process first, spreadsheet second, software third. Every step proves the next one is worth paying for. Businesses that skip to step three don't skip the first two, they just pay developer rates to discover them, mid-project, as change requests.
And when you do arrive at step three legitimately, arrive with your battle-tested spreadsheet under your arm. It's the best briefing document custom development can receive: every column is a requirement, every workaround is a feature request, every breaking point is the scope. You'll get a sharper build, faster, for less, because you paid for the thinking in discipline instead of invoices.
The agencies that lose money on this article aren't us. They're the ones who would have sold you the app in month one.