Reduce the cost of failure and then experiment with confidence. This is also a different version of make change easy access then make the rest change

Plan the metric your will use in advance. Based on the metric you decide whether to double down or ditch it. Start thinking in terms of bets