What 16,000 projects say about your next programme budget
<p>Across a dataset of 16,000 projects, Bent Flyvbjerg found that cost overruns and schedule slippage are closer to the norm than the exception. Most experienced programme leaders already suspect this. The contribution is that someone finally quantified it.</p>
<p>The mechanism is what Kahneman and Tversky called the Planning Fallacy. Leadership teams tend to forecast from the 'inside out', drawing on their own experience, their own team's assumptions, and their own confidence about why this particular project is different. The research suggests this operates even when leaders are aware of it…which is the uncomfortable part.</p>
<p><strong>What does the distribution actually say?</strong></p>
<p>Reference Class Forecasting asks a fundamentally different question: instead of "how long do we think this will take?", it asks "what happened across comparable projects, and what does that distribution tell us?" You find a relevant class of similar efforts, look at what actually occurred, and use it to calibrate your own estimates. It is not a replacement for expert judgement. It is a correction for the tendency of expert judgement to over-rely on the 'inside view'.</p>
<p><strong>The 'Goldilocks data' problem</strong></p>
<p>The practical barrier is data quality. Internal project data is often politically sanitised (budget overruns get reclassified, scope changes explain away delivery failures, post-mortems tell a more flattering story than reality warrants). External data can be too broad or too stale to serve as a genuine reference class. The sweet spot…what the full article calls 'Goldilocks data'…is recent, relevant, honestly reported data from projects that genuinely resemble yours. Most organisations do not have it. Few are building it.</p>
<p>This is the kind of diagnostic conversation <a href="https://app.pragmaticchange.com.au/" target="_blank">Pragma</a> is built for. If you are scoping a complex programme and want to pressure-test your assumptions against 'outside-in' thinking, bring a specific scenario and see what surfaces. It will not replace Flyvbjerg's statistical rigour…but it can surface the assumptions your planning is quietly ignoring.</p>
<p>For the complete analysis, including the practical question to introduce in your next planning session and why the author's Lean Six Sigma colleagues deserve a shout out, read <a href="https://www.markwinter.com.au/thoughts/outsidein" target="_blank">the full article on markwinter.com.au</a>.</p>