The question “custom system or off-the-shelf software” usually appears when a spreadsheet stops being enough and a subscription for a tool you use three features of starts to chafe. The answer does not depend on company size but on how far your process differs from the standard.
Four signs a custom application will pay off
- You enter the same data in several places. Email → spreadsheet → accounting → message to the client. Every retype costs time and creates mistakes.
- Your most important knowledge sits in one spreadsheet on one computer. If its failure stops the company, that is a risk, not a tool.
- You pay for features you never use while the one that matters to you is missing.
- You have devices to connect — cameras, sensors, controllers. Standard systems rarely speak to the exact set you own.
When a custom system is a bad idea
Honestly: in many cases ready-made software wins. Accounting, invoicing, payroll, standard warehouse handling — these are regulated by law and refined over years. Writing them from scratch is expensive, risky and usually produces a worse result than a modest subscription.
| Criterion | Off-the-shelf | Custom application |
|---|---|---|
| Starting cost | Low or none | One-off, depends on scope |
| Recurring cost | Subscription growing with users | Server and domain, care optional |
| Fit to your process | You adapt to the software | The software adapts to you |
| Time to launch | Immediate | Weeks, depending on scope |
| Device integrations | Only what the vendor planned | Whatever is technically possible |
| Data ownership | With the vendor | On your own server |
The first version: smaller than you think
The most common mistake is trying to build the whole system at once. A good first version covers one process end to end — for example service tickets with a status, a deadline and an owner. Nothing more. Real use then shows what is genuinely missing and what nobody ever opened.
What the first version should always include
- Sign-in and roles — who sees what and who can change it.
- Change history — who did what and when.
- Data export to a file, so nothing is locked inside the system.
- A backup running before the first real record is entered.
- Usability in a phone browser if anyone works in the field.
Six questions before you decide
- Which specific process should the system handle, and how many hours a month does it take today?
- How many people will use it, and in what roles?
- What data has to enter it and from where — manually, a spreadsheet, an API, a device?
- What happens if the system is unavailable for an hour? For a day?
- Does a ready-made tool really not do this, or does it just do it differently than you are used to?
- Who on your side owns the decisions and the acceptance?
A custom application is a tool, not a status symbol — it pays off when it removes work that cannot be bought as a subscription. To talk a specific process through and hear an honest answer, see custom applications and panels.