It is Startup Week in Boulder, CO. Tomorrow, Thursday 5/19 at 3:00 pm, I am leading a salon style discussion: Why startups should do experiments.
I recently visited Dan Ariely, Duke University professor, best selling author, and proponent of experimentation in business (read more). In anticipation of the event, I asked him:
Why should companies do experiments?
Though Dan catches me a bit off guard, I agree with his answer. Businesses ought to be interested in understanding causal relationships, but the data they have access to is typically correlational in nature. The distinction is important because understanding when X causes Y (e.g, an increase advertising causes an increase in sales) guides business strategy in ways that are far superior to knowing that X is associated with Y (e.g., advertising is positively associated with sales).
Although a correlational relationship oftentimes appears to be causal, there are two ways that this may not be the true: 1) The relationship may be due to a third variable (e.g., a company acquires a competitor, advertises the acquisition, and now has a larger sales force); the acquisition causes both effects, or 2) the causal relationship may occur in the reverse (e.g., greater sales leads to more money being put in the advertising budget).
Only an experiment can differentiate correlation from causality. Come to the discussion and learn more about these issues tomorrow.