Few organizations set out to run an ERP program that drags on for years. Yet many institutions live with a system that went partly live, a backlog of open issues and teams that have quietly returned to spreadsheets. The software is usually capable of doing the job. What went wrong tends to sit in how the implementation was scoped, owned and run.

The common causes

  • Scope defined by modules rather than processes. Lists of modules say little about how work actually moves between departments, where approvals happen or which records are trusted.
  • Data migration left until the end. When migration starts late, poor-quality data arrives at go-live and users lose confidence in the system immediately.
  • Unclear business ownership. If the ERP is seen as an IT project, decisions about process and policy are delayed or made by default.
  • Customization used to avoid decisions. Every customization made to reproduce an old workaround adds cost, testing effort and future upgrade risk.
  • Training treated as an event. A few sessions before go-live are not enough for people to change how they work.

A structured recovery

Recovering a stalled implementation is less about starting again and more about restoring control. A practical sequence looks like this.

1. Stabilize what is live

Before redesigning anything, fix the issues that block daily operations. Users need to see that reported problems are being resolved; that rebuilds the credibility the rest of the recovery depends on.

2. Assess honestly

Review the configuration, customizations, data quality, open issues and how each department actually works today. The aim is a shared, factual picture: what works, what is being worked around and what was never finished.

3. Re-baseline around processes

Redefine the remaining scope as end-to-end processes, such as procure-to-pay, order-to-cash and record-to-report, with a named business owner for each. Decide which customizations to keep, simplify or retire.

4. Treat data as a workstream

Assign owners to master data, agree cleansing rules and run migration rehearsals. Reliable data is what turns a configured system into a trusted one.

5. Deliver in phases

Plan the remaining work in phases that each deliver visible operational value, with clear acceptance criteria. Smaller releases reduce risk and keep momentum.

The engagement should begin with the operating model rather than the software screen.

Signs the recovery is working

Month-end close becomes more predictable, management reports are produced from the system rather than rebuilt in spreadsheets, open issues trend down and departments stop maintaining parallel records. Those are the measures that matter, more than the number of modules marked as live.