Custom ERP or Off-the-Shelf? An Honest Decision Framework
SAP, Odoo and Zoho are often the right answer. Here is how to tell when they are not, using total cost of ownership rather than feature lists.
We build custom ERP systems, so you might expect this article to argue for custom ERP. It does not. Most companies that ask us for one are better served by a package, and saying so early saves everyone six months.
Here is the framework we actually use.
Start with the licensing arithmetic
Before discussing features, work out the five-year cost of the packaged option:
(seats × per-seat monthly fee × 60)
+ implementation and consulting
+ paid customisations
+ annual supportFor 30 users on a mid-tier plan at roughly $40 per seat per month, that is $72,000 in licensing alone before a single customisation. Suddenly a one-off build is in the same conversation — and the build has no seat ceiling when you grow to 60 users.
For 8 users, the arithmetic goes the other way and it is not close. Buy the package.
Then ask where your advantage lives
Packaged ERP encodes a conventional way of running a business. That is its strength: the processes are already designed and battle-tested.
The question is whether the thing you do better than competitors is one of those conventional processes.
- A distributor whose advantage is buying power and logistics scale → the package models this well. Buy it.
- A manufacturer with an unusual make-to-order configuration process that customers choose them for → forcing that into a package's workflow removes the advantage.
- A microfinance lender with loan products designed around local repayment patterns → no package models this properly.
If your edge lives in a process the package cannot represent, customisation costs will grow every year and upgrades will get harder each time.
The customisation trap
This is where packaged ERP projects go wrong. The pattern is consistent:
- The package covers 80% of requirements out of the box
- The remaining 20% gets handled with customisations and plugins
- Every platform upgrade risks breaking those customisations
- Within three years, upgrades are deferred because they are too risky
- You are now running unsupported software you cannot easily leave
An honest test: ask the vendor how many of your requirements need customisation. If the answer is more than about 20%, you are buying a platform and then rebuilding a meaningful part of it anyway — at a worse cost per feature than building directly.
What custom actually costs
Custom ERP is not just the build. Budget for:
| Cost | Typically overlooked? |
|---|---|
| Initial build | No |
| Data migration from legacy systems | Yes — often the largest line item |
| Staff training and parallel running | Yes |
| Ongoing maintenance (~15–20% of build cost per year) | Yes |
| Hosting and infrastructure | No |
Data migration deserves particular attention. Twelve years of inventory records with inconsistent SKU formats and three different units of measure will take longer to reconcile than the module reading them.
A shortcut: start with one module
You do not have to choose everything at once. The approach that works best is to take the single workflow that hurts most — usually inventory reconciliation or month-end close — and build only that, integrated with whatever you run today.
This gives you:
- Value in weeks rather than a year
- A real measurement of whether custom development suits your organisation
- A working data model to extend, if you continue
- A limited, affordable loss if you decide packaged software was right after all
A decision table
| Situation | Our recommendation |
|---|---|
| Under ~15 users, conventional processes | Packaged (Odoo, Zoho) |
| 15–50 users, one or two unusual processes | Packaged core + custom modules for the unusual parts |
| Competitive advantage in a process no package models | Custom |
| Heavily regulated with specific local reporting | Custom, or packaged with a local compliance partner |
| Already customising a package beyond ~20% | Reassess — you are paying twice |
| No clear internal process owner | Neither yet. Fix that first. |
That last row matters more than any technical consideration. An ERP project without someone internally empowered to decide how a process should work will fail regardless of which software you choose.
If you are weighing this up, we are happy to give you a straight answer — including "buy the package." Start a conversation, or read more about our ERP system development work.