The question almost always comes in the same order: a team loses time on a task, someone suggests 'a small tool,' and the conversation slides into technology choice in thirty seconds. The reverse order leads to good decisions.
Three options exist, and only one is right in a given context: automate what already exists, purchase a market solution, build a custom tool. The costly mistake is not choosing the wrong tool: it’s never having compared the three.
The true cost is not the purchase price
A market software at 40 € per user per month seems less expensive than a specific development. Often, yes, over twelve months. But over five years, the comparison changes, as the real cost includes four unaccounted-for posts.
- The subscription, multiplied by the number of users and years
- The time spent working around what the tool cannot do
- The integrations to maintain between this tool and others
- The exit cost, the day you need to retrieve your data
Conversely, a custom development supports a higher initial cost but a low marginal cost: the tenth user does not cost more than the first. Comparing the two without a time horizon makes no sense.
Four criteria suffice to decide
We do not use any more than necessary. Beyond that, the discussion becomes theoretical and the decision is delayed.
1. Frequency and Stability
A task executed fifty times a month, always the same way, is an obvious candidate for automation. A rare task, or one where rules change every quarter, is not: the maintenance cost will exceed the benefit. The right question is not 'is it automatable?' but 'is it stable enough to remain automated?'
2. The Differentiating Factor
Your pricing method is probably not a competitive advantage: buy. Your way of qualifying a deal, following up on a file, or detecting a signal might be: in that case, a standard tool forces you to fit into everyone else's mold, and you lose precisely what makes you unique.
3. Risk and Dependence
What happens if the publisher doubles its fees, closes its API, or is acquired? If the answer is 'we would be blocked', this risk must be evaluated alongside the price. This does not condemn the purchase: it imposes the need to check data exports and not entrust one's core business to a black box.
4. Operating Costs
Who updates it? Who fixes it when it breaks? Who trains new hires? A custom solution without anyone to maintain it is a liability, not an asset. It's often this criterion, rather than the initial budget, that makes the decision.
The dashboard, all on one page
| Criterion | Automate | Purchase | Build |
|---|---|---|---|
| Frequency of need | High and Regular | Variable | Daily, structured |
| Stability of rules | Strong | Market standard | Tailored to the company |
| Differentiation | Low to medium | None | Real |
| Time to go live | Days to weeks | Immediate | Weeks to months |
| Marginal Cost per User | None | Linear | None |
| External dependencies | Moderate | Strong | Low |
A quick read already gives direction. If two options are still tied, it’s almost always that the problem isn’t precise enough, not that the grid is insufficient.
Three recurring traps
To Monitor
- Build to avoid negotiating. A custom tool launched because a subscription seemed expensive often ends up costing more, without the support.
- Buy to go fast, then bypass. If the team reverts to using spreadsheets three months after the purchase, the tool has not been adopted: it has been tolerated.
- Automate a broken process. Automating a bad procedure only produces the same errors, faster and on a larger scale.
Decide quickly, without rushing into it blindly
The decision consists of five steps, achievable in one to two weeks depending on the availability of the concerned persons.
- Describe the task in one sentence, including its actual frequency and the time it currently consumes.
- Name the expected result and how we will know it has been achieved.
- Test two market tools on a real case, not on a demonstration.
- Quotate the three options over a five-year period, including operations.
- Decide, write the decision, and the date at which it will be reviewed again.
To keep in mind
- Always compare all three options: not comparing them costs more than making a mistake.
- Buy what is standard, build what sets you apart, automate what is stable and repetitive.
- The operational cost, not the initial budget, is the most decisive factor.
- Write the decision and its review date: a just choice today may not remain so indefinitely.