You work on a two-sided ride-sharing marketplace and are considering an experiment that changes how trips are matched or dispatched. Early results look promising on one side of the market, but the team is worried the test may be distorting behavior across riders and drivers.
What are the main pitfalls when running experiments in a two-sided marketplace like Uber, and how would those pitfalls affect whether you trust the result?
You work on a two-sided ride-sharing marketplace and are considering an experiment that changes how trips are matched or dispatched. Early results look promising on one side of the market, but the team is worried the test may be distorting behavior across riders and drivers.
What are the main pitfalls when running experiments in a two-sided marketplace like Uber, and how would those pitfalls affect whether you trust the result?
