Custom software or off-the-shelf: how to choose

There is one sign that is almost always right. If you have already tried two or three off-the-shelf applications and all of them "almost fit", the problem is usually not the applications.

Almost fit means it does not fit — it is only similar. And every application that almost fits asks the same thing of you: that your team changes, not the software.

Start with one question, not a feature list

Do not start with "what features do I need". Start with this: does the way I work make me win, or is my way of working the same as my competitors'?

If your way of working is the same as your competitors', buy off-the-shelf. You are not building an advantage, you are closing a gap — and for that, off-the-shelf is always faster and cheaper.

But if there is one thing in how you work that makes customers choose you — how you price, how you hold stock, how you approve spending — that is exactly where off-the-shelf software will hold you back instead of help.

When off-the-shelf is clearly better

The clearest example: point of sale. Almost every shop needs the same things — scan the item, total the bill, print the receipt, record who sold it. There is no competitive advantage in that.

An off-the-shelf point-of-sale costs roughly one hundred to five hundred thousand rupiah per month. A built-to-order one starts in the tens of millions, plus annual maintenance. The difference is not small — it can be ten times over three years.

The same applies to payroll, attendance, and simple bookkeeping. They all have their own pattern, and thousands of businesses use that same pattern.

When custom is actually cheaper

This is the part most often ignored: custom software is sometimes cheaper, not more expensive.

Take a distributor with three warehouses. Every day, one person copies stock figures from the warehouse system into a spreadsheet and sends it to head office. That work takes two hours. If you have one person costing four million a month who spends half their time on this, you are throwing away roughly twenty-four million a year — every year, without end.

A small system that unifies the three warehouses might cost thirty to fifty million, built once. Within two years it has paid for itself. After that, you keep saving.

Calculations like this are rarely done, because the cost of copying data never appears as its own line in the accounts. It shows up as overtime, as stock errors, as customers disappointed because the item they ordered turned out to be out of stock.

Three numbers to ask for before deciding

Any vendor — off-the-shelf or custom — should be able to answer these three things with numbers, not adjectives.

• How long until the system is usable day to day, counted from the day of agreement, not the day coding starts.
• What the total cost is over three years: build or subscription, plus adjustments, plus maintenance.
• What the exit cost is: if you stop using it or change vendor, how your data comes out and in what form.

The third question is the one least asked, and the one that matters most. Many companies discover that their data cannot be exported in a useful form after three years with one vendor, and that the cost of leaving is larger than the subscription they were trying to avoid.

The middle path rarely offered

The choice does not have to be either one. Use off-the-shelf for the common pattern — point of sale, accounting, payroll. Build the genuinely distinctive part yourself — how you calculate commission, how you route deliveries, how you approve purchases — and connect it to the off-the-shelf system through an API.

This is rarely offered because it does not sell as a tidy package. Yet for many companies it is the cheapest and most durable option: you do not pay to rebuild what already exists, and you are not forced to abandon the way of working that makes you win.

Danger signs in a proposal

If the proposal does not state in writing what you will receive, how long it takes, and what it costs after the system is live — do not sign yet.

If a vendor says everything can be done without first asking how you work, that is not confidence, it is ignorance. Work that begins without understanding ends with repeated scope changes, and a cost that grows without a clear result.