Programmes don't fail by accident. They fail in ways that were written down, published and freely available years before each new failure began. A database of more than 16,000 projects shows that only one in two hundred comes in on budget, on time and delivers the benefits it promised. The reasons are not a mystery. National audit offices, public inquiries and accident boards have been listing the same causes for forty years: unclear accountability, weak sponsorship, clients who can't challenge their suppliers, benefits nobody measured and optimism nobody corrected. So why do organisations full of capable people keep doing the opposite? In The Programme Delusion, Steve Griffin starts with the psychology most delivery books skip. Hindsight bias, uniqueness bias, the illusion of control and the slow drift of standards all let smart leaders believe the statistics apply to someone else. He then sets out ten commandments. They aren't techniques you apply when it's convenient. They are standing commitments meant to hold when the situation, the budget or the politics argue against them. Each commandment is built on real, publicly documented cases, including NASA's Challenger and Columbia, the FBI's Sentinel programme, Heathrow Terminal 5, the Sydney Opera House, Boeing's MCAS, TSB, the Post Office and New Zealand's Novopay. Each chapter: Opens with two closely matched programmes that went in opposite directions Sets out what the evidence actually says, and retires the popular statistics that don't survive scrutiny Names the anti-pattern so you can spot it in your own programme Gives practical tools, each described both as something you produce and as a behaviour you can observe Closes with five uncomfortable diagnostic questions you can ask about your own programme in an afternoon A final chapter looks at what AI changes, and what it doesn't, for programme delivery. For anyone who sponsors, leads, governs, assures or funds programmes and projects, especially at the moment you're about to approve one. Most of the failures in this book were locked in early, at a table where someone could have asked a question and didn't. It all comes down to one question you can ask on any programme, on any given Tuesday: What are we currently choosing not to find out?
