Standish CHAOS Report: Why Agile Projects Still Win

When it comes to technology projects, agile approaches succeed where traditional, plan-driven ones fail. It really is close to that simple. The Standish Group’s CHAOS research — drawn from decades of studying software project success and failure — has consistently found that agile projects are markedly more likely to succeed, and markedly less likely to fail outright, than waterfall ones. In their widely-cited figures, agile projects come out roughly three times more likely to succeed, and waterfall projects around twice as likely to fail.

To be clear, these success and failure metrics are about software projects specifically. I make no claim about other kinds of project.

Oddly, not everyone agrees, even about software. Parts of the project-management community spent years decrying agile ways of working, and only latterly embraced their own version of it — while still insisting traditional approaches have their place on technology initiatives. Unwilling to fully let go, they promote the alchemy of “hybrid agile,” whatever that turns out to mean in practice.

But the debate aside, most people now accept that heavy upfront planning and plan-driven delivery struggle on fast-moving technology work. The evidence that iterative, empirical delivery outperforms big-design-up-front is substantial and long-standing.

The more interesting question for the sectors I work in — defence, aerospace, financial services — is not whether agile delivery wins, but how you get those results when your output is audited. In regulated environments the failure modes are more expensive and the assurance bar is non-negotiable, so the answer is never “just be more agile.” It’s to build an operating model where fast, iterative delivery and rigorous assurance are produced by the same disciplined engineering — small, well-evidenced increments rather than big, infrequent releases. That’s where the CHAOS numbers actually get realised in a world that can’t afford to fail loudly.