Ask someone how long their project will take: they answer by picturing how it will unfold. Ask how long comparable projects took: the number changes, and it is closer to the truth. Almost nobody asks the second question.
In 1993 Kahneman and Lovallo described two ways of looking at a project. From the inside, you run the scenario: the steps, the people, what you will do. From the outside, you file the project into a family of comparable cases and look at how those ended.
The inside view is wrong in the same direction every time: too optimistic, on timelines as on costs. This is not a character flaw, it is a way of reasoning.
The remedy fits in one sentence: before estimating, find which family of cases your project belongs to, and what that family produced.
When the subject allows it, Kapari surfaces a number from its knowledge base: a documented order of magnitude on comparable cases, with its source and its link.
Two rules govern it. It must come from a verified, open, readable source. And it must actually relate to your decision: an automated judge discards off-topic numbers before they reach your screen.
On the first measurements, that filter discarded ten out of twenty-four. A number that is accurate but off-topic does more harm than no number at all.
Internal debates often run on impressions. One person thinks it will be fine, another thinks it will not, and nobody has a point of comparison.
A verified number moves the conversation. You stop discussing what you believe and start discussing what happened elsewhere.
And when no solid number exists, Kapari does not invent one. It shows none, and says so.
Kahneman, D. and Lovallo, D. (1993), "Timid Choices and Bold Forecasts: A Cognitive Perspective on Risk Taking", Management Science. On the same family of biases applied to megaprojects: Flyvbjerg, B. (2006), "From Nobel Prize to Project Management: Getting Risks Right".
Kapari measures its audience with Google Analytics, to know how many people read these pages. Nothing is stored on your device until you accept.