ETA Experiment

Let’s say you work at Uber. A PM comes to you considering a new feature where instead of a direct ETA estimate like 5 minutes, would instead display a range of something like 3-7 minutes.

How would you conduct this experiment and how would you know if your results were significant?