7 ERP Implementation Mistakes That Quietly Kill the Project

I’ve sat on both sides of an ERP project — writing the code as the engineer, and owning the delivery as the manager who has to answer for the result. After enough of them, you stop blaming the software. When an ERP project goes sideways, the tool is almost never the real reason. The damage was usually done months earlier, in a handful of decisions that looked perfectly reasonable at the time.

Here are the seven I run into most, and what I do instead.

1. Trying to launch everything at once

The most expensive mistake is the “big bang” — sales, purchasing, inventory, production, HR, and finance all going live on the same day. Every extra module multiplies what you have to test, what you have to train people on, and what can break when the system is finally switched on.

Pick the one or two areas that hurt the most right now — usually sales and inventory — and ship those first. A tight first phase earns trust, proves the thing works, and gives your team something real to learn on before you pile on the rest.

2. Carrying your mess into the new system

Dirty data doesn’t disappear in a migration. It just shows up again, at scale, with shiny new reports built on top of it. Duplicate customers, product codes that don’t match, half-finished records — all of it follows you across, and the moment people spot it, they stop trusting the system.

So clean it first, as its own piece of work, before anyone migrates anything. It’s dull, nobody thanks you for it, and it’s probably the highest-leverage thing you’ll do on the whole project.

3. Customizing too much, too early

Each customization feels justified on its own. Add up enough of them and you’ve built a fragile, one-of-a-kind system that costs a fortune to maintain and fights you every time you try to upgrade. Bend your process to fit the standard workflow wherever it’s good enough, and save customization for the few places where you’re genuinely different — and where that difference is worth carrying forever.

4. Leaving compliance until the end

In Saudi Arabia, that means ZATCA Phase 2 e-invoicing. Treat it as a last-minute bolt-on and you’ll find out the hard way that it touches invoicing, POS, and reporting — basically the heart of the system. Design it into the invoicing flow from day one. Bolting it on under deadline pressure is exactly how a clean go-live turns into a fire drill.

5. Nobody actually owns it

When everyone owns the project, no one does. Decisions stall, scope quietly drifts, and your implementation partner ends up making business calls they were never supposed to make. Put one person in charge — someone with the authority to decide and the responsibility for how it turns out. Clear scope, a decision log, managed risk: this is what PMP discipline is for.

6. Forgetting that people have to use it

I learned this one on the Mawan rollout. The hard part was never the technology — it was building something the team actually wanted to open every morning. A technically flawless system that everyone quietly works around is, in every way that matters, a failed system. Bring the people who’ll use it in early, train them on the real workflow, and treat adoption as a number you track — not something you cross your fingers about.

7. No real definition of “done”

“Modernize our operations” isn’t a target. You can’t hit it, miss it, or defend the budget you spent chasing it. Decide what success looks like in numbers before you start. On Mawan the targets were specific, and so were the results: order processing 40% faster, production efficiency up 30%, and the backbone that carried 120% revenue growth.

The thread running through all of it

Six of these seven have nothing to do with code. ERP is a business-and-people project that happens to involve software. Get the scope, the data, the ownership, and the adoption right, and the technology mostly takes care of itself.

If you’re planning an ERP project — Odoo, custom, or a bit of both — and want a second opinion before you commit, let’s talk. A short conversation now is a lot cheaper than a stalled rollout later.