In the IPO prospectus for Google, the company stated that it would invest in projects that had a 10% chance of success if the return could be a billion dollars plus. The company also said it would ‘place smaller bets in areas that seem speculative or even strange’. In other words, Google will invest in strategies and products that are more likely to fail than succeed.Failing often, failing fast and failing cheap (otherwise known as experimentation) will be a core competence in the future, but how do you create a culture that not only accepts and learns from its mistakes but does things that it knows will fail? The difference is important. Unlike scientists, managers do not deliberately try to prove themselves wrong. Instead, they use historical evidence to prove that something will work in the future. Scientists, work on a different basis – that you can never prove something is right, only that it is wrong. In other words, only through making mistakes can you disprove a hypothesis and the more mistakes you make the faster you will disprove it. So how can you harness the power of deliberate mistakes to improve your strategy or innovation process? Or to put it another way, how can you tell the difference between a smart mistake and a dumb one? To some extent the answer is not about how but when. Mistakes made on purpose are best used when the potential gain of being right far outweighs the cost of being wrong. Another reason to make deliberate errors and study the outcome is when rigid assumptions drive large numbers of small decisions. For example, it was once a core assumption that giving credit cards to students was a bad idea. However, Citibank decided to test that assumption and found that that the ‘worst possible risks’ were in fact nothing of the sort. Other instances when making deliberate mistakes can pay off include situations where the industry landscape is changing fast, where problems are complex and solutions are numerous or where experience with a situation is limited. In each of these cases deliberately making a mistake and learning from it can be a powerful strategy but it is rarely something that large organisations feel comfortable with. Programs like Six-Sigma have created measurement systems and cultures that aim to eliminate failure, not promote it. Employees feel insecure because of short-term employment contracts and evaluations so while they buy into ‘mistakes done right’ in theory, in practice they are loathe to put their own careers on the line. One solution to this dilemma is to evaluate employees on longer timeframes.This is something that IBM Research is trying. Bonus payments are linked to one-year performance, but salary and rank are linked to three-year timeframes. Maybe they’ve been listening to Thomas Watson, the founder of IBM, who once said ‘ If you want to succeed, double your failure rate’.