Insights
The famous statistic says most ERP programmes fail. The famous statistic is also mostly quoted by people selling the cure. Here is what actually goes wrong, written by someone who gets called when it has.
By Christopher Fagan, PEF Consulting Partners. 3 September 2026.
You will read that fifty, sixty or seventy percent of ERP implementations fail. Treat those numbers with suspicion: they come from surveys with shifting definitions of failure, and they are usually cited to sell you something. What is true from delivery experience is simpler and more useful: outright cancellations are rare, but programmes that run over time, over budget and under scope are the norm rather than the exception, and the causes repeat with remarkable consistency.
The plan was never real. It was built to win the approval or win the contract, priced on optimism, with the hard questions deferred. Everyone senior enough to know better signed it anyway, and the programme spent the next two years discovering the truth one slipped milestone at a time.
The client hollowed out its own side. Design decisions, data ownership and delivery assurance were left to the integrator, and an integrator resolves every ambiguity in the direction of its own commercial interest. Not from malice, from incentives. A client without senior capability on its own side of the table has outsourced its judgement along with its build.
Data started too late. Data work is unglamorous and demo screens are not, so the data workstream starts second and finishes last. Then the first full trial load arrives and the programme discovers what its estate actually contains, at the point where every fix is on the critical path.
Scope grew to recreate the past. Every legacy exception, workaround and side report gets rebuilt in the new system, because change is uncomfortable and saying no takes authority. The result is an expensive new platform running the old operating model, delivering the worst of both.
Governance became theatre. Meetings that receive status instead of making decisions, attended by people who cannot commit their function, fed by reports that reward green. By the time the board hears the truth, it has been true for six months.
They invest before the clock starts: readiness, data and decision rights sorted before the integrator arrives. They keep senior, independent capability on their own side for the whole journey. They treat standard process as the default and customisation as an exception needing a business case. They protect testing and operational readiness when dates get tight, because those are the workstreams that decide whether go live is an event or an incident. And they reward the first person to bring bad news, which keeps the reporting connected to reality.
Failure is rarely sudden and rarely final. The early signals, slipping milestones that never shrink, compressed testing, a silent programme, are covered in our guide on how to rescue a failing ERP programme, and our free health check gives you an honest read in ten minutes.
Either way, the free tools are the place to start. No sign up, answers stay in your browser.
Related reading: what ERP really costs and how long it really takes.