How to rescue a failed ERP implementation
A stalled ERP is not a write-off. Most failed projects are recoverable, and the recovery usually costs less than the original mistake. Here is the framework we use.
The call usually comes twelve to eighteen months in. The system is 'live' but half the company still works in Excel. Or it never went live at all, and the invoices keep coming. Nobody wants to touch it because the last attempt burned money and credibility. The good news, from the projects we have recovered across the Gulf: the software is almost never the problem, and most of the sunk cost is salvageable.
Key takeaways
- Diagnose before deciding: most 'failed ERPs' are failed implementations of workable software.
- The keep-fix-restart decision follows from root cause, not from frustration.
- Recovery starts with one workflow made genuinely excellent, not a relaunch of everything.
- Whatever broke the first project (unscoped work, hourly billing, no gates) must be structurally impossible in the rescue.
Step 1: triage honestly
Three questions determine everything. What was actually delivered against what was promised? Where does work really happen today, in the system or around it? And is the data in the system trustworthy? Write the answers down; the rescue plan falls out of them.
Step 2: find the real root cause
In our experience the causes rank like this: scope was never written down, so the project could not finish; delivery was billed hourly, so there was no incentive to finish; migration was never validated, so users never trusted the numbers; training was a handover PDF, so adoption died; and only rarely, the platform genuinely did not fit. Note that four of the five are contract and discipline problems. That is why they are fixable.
Step 3: keep, fix or restart
| Situation | Call |
|---|---|
| Right platform, bad configuration | Fix in place: re-blueprint, reconfigure, remigrate |
| Right platform, dead adoption | Fix the workflows users bypass, then retrain role by role |
| Untrusted data, decent config | Controlled remigration with signed reconciliation |
| Platform genuinely misfit | Restart, but salvage the process knowledge and cleaned data |
Step 4: relaunch small and win
Pick the one workflow causing the most daily pain, usually sales orders or purchasing, and make it excellent: configured right, data verified, users trained, old workaround switched off. One visible win rebuilds the credibility that the first project spent. Then sequence the rest through gated milestones, each with sign-off, exactly as a healthy project would have run from the start.
Make the old failure impossible
Whatever the root cause was, the rescue contract must preclude it: written blueprint, fixed price, senior delivery, validated migration, gates you sign. If a partner offers to rescue your project on open-ended hourly billing, they are offering you the original problem with a new logo.
Is your ERP project stuck?
Rescues are some of our best work. We audit what went wrong, design the remediation plan, and price the recovery fixed before a line of code is touched.
Book a discovery call →