For the past 3 weeks I’ve eaten a bacon sandwich every Friday morning. So, presumably, it’d be reasonable to assume that I’m going to eat another bacon sandwich this coming Friday morning too. I certainly hope so. I like bacon sandwiches.
This is an example of inductive reasoning – that events in the past point the way to what will happen in the future. Not that they definitely will, but you can certainly infer the probability of something happening based on past data. The important thing to note though, as the philosopher David Hume pointed out in the mid-19th century, is that you cannot say for certain that past events inform us about future events. Things change.
Hume’s “Principle of the Uniformity of Nature” argues that we have a tendency to believe things happen with uniform regularity, that patterns we recognise will continue in to the future. Experience should tell us that this is wrong. We have plenty of information demonstrating how things change but we ignore it.
Start-ups are often guilty of inductive reasoning. Arguments that, for example, “all our early adopters liked feature X, therefore it will be important to all the users” are commonplace. Inducing knowledge by making assumptions is a critical failure that can lead your business down entirely wrong paths.
200 years later this principle is one of the key ideas behind Eric Ries’ Lean Startup movement – “Actionable Metrics”. Actionable metrics are things that can be used to drive decisions about what to do next based on repeatable, rigorous tests. Rather than trusting assumptions or deciding things based on historical data points, we should be testing whether or not things have been true in the past and will remain true in the future before deciding what to do. Actionable metrics are based on deductive reasoning.
Deduce knowledge from what you can prove, don’t induce knowledge from what you think you know.