When programmes go wrong

The odds are worse than boards assume

Most boards know large programmes are risky. Few know how risky. Bent Flyvbjerg's database at Oxford — the largest of its kind, covering more than 16,000 major projects — shows that just 8.5% come in on budget and on time. Add "delivers the promised benefits" and the figure falls to 0.5%. One project in two hundred.¹

Technology programmes are no exception. McKinsey and Oxford's study of 5,400 large IT projects found they run, on average, 45% over budget and deliver 56% less value than promised. One in six becomes what the authors call a black swan — an overrun so severe it threatens the existence of the company itself.²Transformation programmes fare no better: BCG has reported for years that roughly 70% of transformations miss their targets, a figure that has barely moved despite two decades of method and tooling.³ PMI estimates that around a tenth of every invested dollar is simply wasted through poor project performance.⁴

Behind each of these numbers is the same uncomfortable truth: the organisation almost always knew earlier. Not officially — but somebody knew.

Programmes fail slowly, then suddenly

A critical programme rarely collapses without warning. It decays — and the decay is systematically hidden by four mechanisms that operate in almost every large organisation.

Optimism at the start

Budgets and timelines are set when knowledge is lowest and enthusiasm is highest. Flyvbjerg calls the result the planning fallacy at scale: baselines that were never achievable, defended long after the evidence has turned.¹

Watermelon reporting

Green on the outside, red on the inside. Each management layer softens the message slightly, for understandable human reasons — and by the time status reaches the board, a programme in serious trouble reads as "amber, mitigations in place."

Sunk-cost escalation

The more a company has spent, the harder stopping becomes. Decisions stop being about the next dollar and start being about justifying the last ten million. This is how one bad year becomes three.

Governance decay

Steering committees grow, decision rights blur, and meetings drift from decisions to status theatre. BCG's work on large programmes points to governance — not technical complexity — as the dominant driver of failure.³

Eight signals a programme is in trouble

None of these requires technical insight. All of them are visible from the boardroom — if you know what to look for.

1. 01

   The baseline keeps moving.

   A programme that replans every quarter does not have a plan. It has a negotiation.

2. 02

   Status is green, but nothing ships.

   Milestones pass, demos slip, and 'done' keeps being redefined. Trust delivery, not decks.

3. 03

   Contingency is quietly gone.

   Reserves absorbed early in a programme's life are a statement about the estimate, not about bad luck.

4. 04

   Key people are leaving.

   The engineers and managers closest to the truth leave first. Exit interviews know more than status reports.

5. 05

   Scope is being 'deferred'.

   Descoping without a decision is failure on an instalment plan — the benefits case erodes one deferral at a time.

6. 06

   Dependencies slip without consequence.

   When missed hand-offs stop triggering escalation, the plan has already stopped being real.

7. 07

   Testing is being squeezed.

   A fixed end date with a slipping build always takes its time from quality. That debt is repaid publicly, after go-live.

8. 08

   Bad news has stopped arriving.

   Healthy programmes surface problems constantly. Silence is not stability — it is fear.

Three or more of these, sustained over two reporting cycles, is not a rough patch. It is a pattern — and patterns do not fix themselves.

The first 30 days of a rescue

When a board accepts that a programme is in trouble, the instinct is to demand a recovery plan from the people who produced the current one. That is usually a mistake. What a troubled programme needs first is not a new plan — it is an honest picture, built fast, by someone with no stake in the story so far.

Days 1–10: Rapid assessment

A short, structured diagnosis of the essentials: what has actually been delivered against what has been reported; whether the remaining scope, budget and dates are connected to reality; whether the benefits case still holds at the true cost to complete; and what the people closest to the work — not the steering committee — say when asked directly. Ten days is enough, because the goal is decision-grade truth, not a perfect audit.

Days 10–20: The honest decision

The assessment feeds one deliberately uncomfortable question: knowing what we now know, would we start this programme today? If yes — recover it. If yes, but not in this shape — reset it: rebase the scope, the plan and the governance around what is true, not what was promised. If no — stop, and redirect what remains of the budget to something that can deliver. All three outcomes are legitimate. Treating only "recover" as acceptable is how companies buy the same failure twice.

Days 20–30: Reset the foundations

Whichever path is chosen, the same foundations follow: one accountable owner instead of a crowded steering committee; a baseline the delivery team actually believes; contingency that is visible and owned; reporting rebuilt around delivered outcomes rather than activity; and an explicit agreement about what bad news is for — steering, not blame. Programmes do not fail because problems exist. They fail because problems stay hidden.

What boards should ask

Five questions, asked calmly and regularly, surface more truth than most assurance frameworks:

• — What has been delivered — not built, delivered — this quarter?
• — If we started today, would we approve this programme in its current shape?
• — How much contingency remains, and who decided how it was spent?
• — What is the worst news inside the programme right now, and who told us?
• — What would have to be true for us to stop?

The last one matters most. A programme with no stopping conditions is not a commitment. It is a blank cheque.

Stopping is a result

The hardest rescues are the ones that end with a recommendation not to continue. They are also, sometimes, the most valuable. The money already spent is gone regardless of the decision. The only question that matters is where the next dollar goes — into a programme that can deliver, or into the defence of one that cannot. Boards that can make that call early, on evidence, turn their worst programmes into their cheapest lessons.

About the author

Robert Kalbarczyk is the partner of Pachira US LLC— a strategic advisory firm working with owners, boards and investors. His operating career spans CEO, COO and board roles across e-commerce, banking, energy, telecom and the public sector, including six years leading operations, strategy and change at a global bank. Rescuing troubled programmes is one of Pachira's four practices.

The Critical Moments practice