We build custom ERP systems, and we still tell most people asking about one to buy a package instead. The cases where building wins are real but narrower than the enthusiasm around them suggests.
Buy, in most cases
Standard ERP platforms encode decades of accumulated process knowledge across thousands of businesses. If your operation is conventional — you buy things, make or move things, sell things, and account for it — that knowledge is worth far more than the flexibility of a custom build.
You also get an ecosystem: consultants who know the product, staff you can hire who already use it, documentation, and a vendor whose business depends on keeping it working.
The customisation trap
The pattern that costs the most money is not choosing wrong. It is choosing a package and then customising it heavily to fit a process it was not designed for.
- Customisation costs frequently exceed the licence cost.
- Heavy customisation makes upgrades painful, so you fall behind on versions.
- You end up maintaining custom code without owning the platform it sits on.
- The vendor's support becomes progressively less useful as you diverge.
When building genuinely wins
- 1Your process is a competitive advantage. If how you operate is why customers choose you, generic software will flatten exactly that difference.
- 2You need a handful of modules, not a suite. Paying enterprise licensing for functionality you will never switch on is a poor trade.
- 3Your industry is niche enough that no package fits. Specialised manufacturing, regulated workflows, and unusual service models often have no good off-the-shelf answer.
- 4You are keeping your accounting system. Modernising operations while finance stays put is often better served by a custom operational layer that integrates with the ledger.
| Your situation | Lean towards | Why |
|---|---|---|
| A conventional operation: you buy, make or move, sell, and account for it | Buy a package | Decades of process knowledge, plus consultants, staff and a vendor who know the product |
| The package needs heavy customisation to fit | Compare it with a focused build | Customisation often exceeds the licence cost and makes upgrades painful |
| Your process is why customers choose you | Build | Generic software flattens exactly that difference |
| You need a handful of modules, not a suite | Build | Enterprise licensing for functionality you never switch on is a poor trade |
| A niche or regulated industry with no good package | Build | Specialised manufacturing and regulated workflows often have no off-the-shelf answer |
| Your accounting system works and your auditors know it | Keep it, and build the operational layer | Statutory accounting changes constantly and carries compliance risk |
The hybrid most people should consider
Keep your accounting software. Almost nobody should rebuild statutory accounting — it changes constantly, carries compliance risk, and your team and auditors already know the current system.
Then build custom only for the operational layer that is genuinely specific to you, and integrate it with the ledger. You get the fit where fit matters and the stability where stability matters.
If you do build
Build module by module. The reason ERP projects have their reputation is that they are attempted as one enormous release, so nothing is usable until everything is finished and the requirements have moved by then.
Get one module — usually inventory or procurement — into production within three or four months. Learn from real usage. Then build the next one against what you have actually learned rather than what you assumed a year earlier.