Before scaling an innovation it is necessary to have quantitative evidence that supports our initial hypotheses.
We should carefully deal with the biases problem to estimate a correct impact: we must isolate it from the external factors and self-selection. To do this, we use three statistical methods: RCTs (Randomized Controlled Trials), the DID methodology (Differences in Differences) and Matching methods.
We recommend interpreting the results with a Bayesian analysis to make business decisions, since it is possible to directly calculate the probabilities of different scenarios.
Inventures measurement support was key to isolate the effect of the pilot from other factors.Tomás Durandeau - Corporate Marketing Manager SMU