The short answer
Every platform embeds assumptions about how you work. If those assumptions are right and you are missing features, configure or extend. If the assumptions are wrong, no amount of configuration fixes it and you will spend years fighting the model.
Feature lists distract from this. The question is about the underlying model, not the checkbox.
Test the model, not the features
- Does it assume one customer per order when you have several?
- Does it assume a linear approval when yours branches?
- Does it assume one price per product when yours varies by contract?
- Does it assume stock in one place when you hold it in five?
- Does it assume a unit of work that matches yours?
A no to any of these is a structural mismatch. You can usually work around one. Three or more and you are building a system inside someone else's product, at a disadvantage.
The honest comparison
| Configure | Build | |
|---|---|---|
| Time to first value | Weeks | Months |
| Fit to your process | Constrained by the model | Whatever you specify |
| Ongoing maintenance | Vendor, mostly | Yours |
| Upgrade path | Provided, may break customisations | Yours to manage |
| Cost over five years | Licences plus workarounds | Build plus maintenance |
| Risk if the vendor changes direction | Real | None |
| Risk of building the wrong thing | Lower | Real |
Neither column is better. The bottom two rows are the ones people leave out, and they point in opposite directions.
Costs each side forgets
Configuration hides cost in workarounds, manual steps and licence growth as you add users. Building hides cost in maintenance, hosting and the ongoing need for someone who understands it.
Compare over five years with both sets included. Over one year, configuring almost always wins. Over five, it depends on how much the workarounds cost you.
The option in the middle
- Keep the platform for what it does well.
- Build only the part where the model does not fit.
- Connect them through the platform's API.
- Keep the platform as the system of record where you can.
- Revisit only if the custom part keeps growing.
This is the answer more often than either extreme, and it is the one that gets skipped because it is less decisive.
How to decide
Write down the three things your business does differently from everyone else in your sector. Those are where custom might be justified. Everything else is almost certainly better bought.
If you cannot name three, that is a strong signal that a platform will fit, and that the frustration is with configuration rather than with the model.