ERP Project Recovery

Rescue or reimplement an ERP: how to decide

When an ERP project has gone wrong, the question isn't whether it hurts but whether to rescue or reimplement it. Criteria for deciding with data.

· 4 min read

After acknowledging that an ERP project has stalled or gone wrong, the question that is truly hard to answer arrives: do you fix what you have, or start over? The answer is almost never obvious, and getting it wrong is expensive in either direction.

This article gives you criteria for deciding with data, not with the instinct of someone who has spent months frustrated with the system.

It is a decision best made once and made well. Changing course midway, starting a rescue and discovering three months later that a reimplementation was needed after all, costs more than making the right decision at the start with all the information on the table.

What each option means in time and cost

Rescuing means correcting what already exists: adjusting configuration, cleaning data, closing processes that were left undefined, training the team that never received complete training. It is measured in weeks or a few months, because it starts from a base that already works in some respects.

Reimplementing means redoing the design from the start, keeping what is useful and discarding what is not. It is measured in months, similar to a new implementation, because in practice that is what it is. It usually costs more than a rescue, and it also solves problems a rescue cannot touch.

The wrong choice in either direction is costly: rescuing a system that needed a complete redesign only postpones the same problem, and reimplementing a system that actually worked spends time and budget solving something that wasn't the cause.

There is a third cost that is rarely counted: the fatigue of a team that has already lived through a bad implementation. Every new round, whether rescue or reimplementation, relies on the same people who will operate the system, and they arrive with less patience and more skepticism than in the original project. Choosing well the first time protects that too.

Signs it makes sense to rescue what you have

A rescue is usually the right answer when the system fundamentally works and the problem is elsewhere.

  • The system correctly processes the operations that were configured; the problem is the modules or processes left half-finished.
  • The migrated data is reasonably reliable, with specific exceptions that can be identified and corrected.
  • The team never received complete training and compensates with spreadsheets for what the system could already do.
  • The original design did match the operation, but the project's execution fell short.

Signs it makes sense to reimplement from scratch

Reimplementation comes in when the problem is not execution but design: what was built does not match how the business actually operates.

  • The configured processes don't reflect how the company works today, or how it worked when the project started.
  • The base structure, such as the chart of accounts or the definition of companies and countries, has fundamental errors that can't be fixed with point adjustments.
  • The data migration was done wrong from the start and has to be redone, not just cleaned up.
  • The system can't handle the current volume or complexity of the operation, regardless of how it is configured.

When the answer is mixed

In practice, few projects fall cleanly into a single category. It is common for one module to be well designed while another needs to be redone from the ground up, or for the process to be right but the data feeding it to be so badly migrated that it should be treated as a reimplementation for that part only.

When this happens, the decision is not to rescue everything or reimplement everything. It is to split the project by module or by process, and apply to each the treatment its own assessment calls for, even if that means running at two different paces within the same project for a while.

What an assessment that answers this question with data looks like

The way to avoid deciding blind is to separate process, data and system as three distinct causes, instead of lumping them into a single feeling that the ERP doesn't work. An assessment reviews each one separately: how much of the configured process does match the operation, how clean the data is, and whether the system itself is capable of supporting what the company needs.

That same document later serves to justify the decision to whoever approves the budget, whether the CEO or a committee. It is different to ask for resources by saying “the system doesn't work” than to arrive with a table showing, module by module, which one is rescued and which is redone, and why.

With those three fronts evaluated separately, the choice between rescue and reimplementation stops being a bet and becomes a calculation: how much each path costs, how long it takes, and what risk remains if you choose the shorter one.

That calculation also serves something harder to accept in the moment: sometimes the assessment shows that neither path is justified yet, and that it makes sense to sustain the system as it is, with targeted support, while the company deals with other priorities. It is not the most common answer, but it is a valid one, and an honest assessment leaves it open until the end instead of closing it off in advance.

If you are at that decision point, the rescue self-assessment gives you a first read in minutes, with ten questions about the state of the project.

Facing something similar?

Tell us which system you run and what it can't do for you today. The first conversation is about understanding your situation.

Request an assessment
Cover of the CETERiS guide From Excel to ERP (in Spanish)

Free guide · 24 pages · In Spanish

From Excel to ERP: how to tell if it's time

What we'd tell you in a first assessment meeting, in writing: how to know whether you need an ERP, what it costs in estimated ranges and how to migrate without disrupting operations. Currently available in Spanish; English edition coming soon.

  • The 7 signs you've outgrown spreadsheets
  • What an ERP costs and what drives the price
  • The 5 most expensive mistakes
  • A checklist to decide whether it's time
Download the guide (Spanish)Free. Delivered to your inbox.