Insights
Most ERP programmes do not fail at go live. They fail quietly, months earlier, while the status reports are still green. Here is how to recognise a programme in trouble and how to take back control, from someone who has recovered them at the largest scale.
By Christopher Fagan, PEF Consulting Partners. 3 September 2026.
By the time an ERP programme is formally declared red, the board has usually been the last to know for six months. The reliable early signals are not in the RAG report. They are these: milestones that keep moving but never shrink, a plan that is replanned more often than it is delivered, testing that keeps being compressed to protect a date, integration and data workstreams that are always about to catch up, and a systems integrator whose change requests are growing faster than their output. Any two of those together deserve an independent look. All five together mean the plan you are reporting against no longer exists.
The most dangerous signal is cultural. When the programme stops bringing bad news to governance, it is not because there is none. It is because the programme has learned that bad news is punished and dates are sacred. That is the point at which the reporting decouples from reality, and it is almost always the point of no return for the original plan.
Having recovered large programmes, including a struggling multi billion pound rollout for BP that spanned five countries, and having owned a global ERP and Microsoft Dynamics 365 estate worth over two hundred million pounds, I can tell you the technology is rarely the root cause. The same four causes appear again and again.
First, the plan was never real. It was built to win approval or to win the contract, priced on optimism, and everyone senior enough to know better signed it anyway. Second, the client hollowed out its own side. Decisions, data and design authority were left to the integrator, and an integrator will always resolve ambiguity in the direction of its own commercial interest. Third, the data work started too late, because data is unglamorous and demonstrable screens are not. Fourth, governance became theatre: meetings that receive status rather than make decisions, attended by people who cannot commit their own function.
Notice that every one of those is within the client’s power to fix. That is the good news hiding inside most failing programmes.
Stabilise before you replan. The instinct in a troubled programme is to produce a new plan quickly. Resist it. First stop the bleeding: freeze scope, pause commitments you cannot honour, and protect the people doing real work from the storm above them. A recovery that starts with another optimistic plan is just the same failure with a later date.
Establish the honest baseline. Independently, and quickly, establish what is actually built, what actually works, what the data really looks like, and what the remaining work truly costs. This is where independence matters most, because everyone inside the programme has a version of the truth shaped by their own exposure. On the BP recovery, the turning point was not a new plan. It was the first honest account of where the programme actually stood, put in front of people senior enough to act on it.
Cut scope like you mean it. Almost every recovered programme goes live with less than the original scope, and almost every one is better for it. Separate what the business needs on day one from what was promised in a workshop two years ago. A smaller go live that lands builds the credibility, and the operational stability, to deliver the rest.
Reset the commercial relationship. A failing programme usually has a failing contract underneath it: misaligned incentives, disputed change requests, and a supplier relationship conducted through commercial letters. Renegotiate around the recovery plan, tie payment to outcomes you can verify, and be willing to change partners where trust is genuinely gone. On the BP programme, resetting the delivery model and supplier arrangements saved twenty five million pounds while the programme accelerated.
Rebuild governance that decides. Replace status theatre with a small group that meets on real evidence, makes decisions in the room, and includes people who can commit money, people and policy. Restore the flow of bad news by rewarding the first person to bring it.
Not every programme deserves rescue. If the business case no longer holds, if the platform choice was wrong for the operating model, or if the organisation cannot resource its own side even after the reset, the courageous decision is to stop, salvage what has value, and rethink. A rescue exists to protect the outcome, not the sunk cost, and an honest assessment will tell you within weeks which side of that line you are on.
Our free Programme Recovery Health Check gives you an honest read in ten minutes, in your browser, with nothing sent to us. Or talk directly to the people who have done this before. Confidential, no obligation.