ERP projects rarely fail at the software. Here is where they actually go wrong

I have sat in enough post-mortems to know how the conversation starts. Somebody puts the vendor on the screen and asks what went wrong with the product. Almost every time, the answer was decided months earlier — in a steering meeting that kept slipping, in a stock master nobody owned, in a go-live date picked to suit a board calendar. The software is usually the least interesting part of the failure.

So when a client asks us to look at a stalled programme, we do not start with the system. We start with the seven things that quietly break ERP in this region, and we score them honestly.

Sponsorship is the first and the most expensive. When a programme is handed to IT or finance and left there, cross-departmental decisions have nowhere to go. Everyone waits for the monthly slot, and the implementer bills for the waiting. The fix is unglamorous: one named executive with the authority to settle a process dispute on the spot, and a fortnightly steering committee working off a live risk register.

Dirty data is the second, and it is the one people underestimate most. Customers, suppliers and stock sitting in four spreadsheets with four different codes. Here is the uncomfortable truth — messy data migrates faster than clean data. It goes in easily and comes back out as distrust in every report after go-live. Clean and de-duplicate the masters first, with a named owner per domain and a signed-off cut-off date. Nothing else in the plan survives if this is skipped.

Scope creep never arrives as a bad idea. It arrives as one more reasonable report, department by department, until the timeline stretches and the original business case has quietly disappeared. We freeze a phase-one scope tied to that case and send everything else to a phase-two backlog with its own approval. Saying “yes, in phase two” keeps the programme honest without making enemies.

Then there is the finger-pointing. Software vendor, implementer, internal team — three credible explanations for the same delay, and no one owning the outcome. The client absorbs both the slip and the rework. This is precisely why we hold no licence revenue: one accountability matrix, contractual milestones, and an advisor whose recommendation follows the evidence rather than the margin.

Timelines are where optimism gets expensive. A date chosen for a financial year, then work compressed to fit it — and testing, training and data cleansing are always the first things cut. We plan backwards from a tested cut-over, not forwards from a wish, and we treat UAT and training as non-negotiable line items.

Change management gets the same treatment. Train people the week before go-live and you will watch the parallel spreadsheets reappear the week after. The system goes live; the people do not. Role-based training, super-users in every department, and adoption tracked as a KPI for ninety days.

The last one is the quietest. The project closes at go-live and nobody checks whether the promised benefits ever arrived — while licences and support renew for years against a business case no one validated. We audit benefits realisation at three and six months, and we act on the gap.

Our consulting sequence exists for exactly this reason. Audit first, then a readiness score, then gap analysis, then structured vendor selection, then implementation governance, then the post-go-live audit — six stages, each one closing a specific pitfall rather than adding process for its own sake.

None of this is theory and none of it is expensive at the right moment. Every one of these problems costs least before contracts are signed and data is migrated. If you are somewhere in that window, walk your programme through the ERP Pitfalls Guide and score it green, amber or red. Optimism is the most expensive input in an ERP budget — and it is the one thing no vendor will flag for you.