The short answer
Pick technology with a large pool of developers, an active community and a predictable upgrade path. Exotic choices are justified by specific requirements, not by preference, and the person making the choice is rarely the one maintaining it in five years.
The question to ask is: if everyone who built this left, how hard would it be to find a replacement?
What actually matters over time
| Factor | Why it matters in year three |
|---|---|
| Size of the developer pool | Determines whether you can hire at all |
| Community and documentation | Determines how fast problems get solved |
| Upgrade path | Determines whether you can stay current |
| Stability of the ecosystem | Determines how often things break beneath you |
| Your team's existing knowledge | Determines the cost of the first year |
Raw performance rarely appears on that list, because for most business software it is not the constraint. Where it genuinely is, that is a specific reason and it justifies a specific choice.
Boring is a feature
Widely used technology has more answered questions, more libraries, more people who have hit your problem, and more candidates when you need to hire. Those compound.
- Problems you hit have usually been hit before
- Libraries exist for the unglamorous parts
- Security issues are found and patched by many eyes
- Hiring is possible at a sensible rate
- Another supplier can pick it up if the relationship ends
When exotic is justified
A genuine requirement that mainstream tools cannot meet: extreme throughput, hard real-time constraints, a domain where one ecosystem dominates, or an existing team who are expert in something unusual.
Those are real. What is not a reason is that the framework is new, interesting, or what a developer wants on their profile.
Questions worth asking a supplier
- Why this rather than the more common option?
- How many developers in our market could maintain it?
- What is the upgrade path, and what happened at the last major version?
- If we moved to another supplier, how hard would it be for them?
- What would you choose if you were not going to maintain it?
The last question is the useful one. A supplier answering it honestly is thinking about your position rather than their convenience.
Consistency beats optimality
A system built in one mainstream stack is easier to maintain than one assembled from the best tool for each part. Every additional technology is another thing to upgrade, secure and hire for.
Where a second technology genuinely earns its place, that is fine. The default should be to use what you already have.