THE HAPPYORNOT BUTTONS—-AN EXPERIMENT WITHOUT ANY CONTROLS. David Owen has an article in the New Yorker (February 5) about HappyOrNot, a Finnish market research company which measures customer satisfaction by means of four push buttons. The buttons are marked “very happy”, “happy”, “pretty unhappy”, and “very unhappy”. The terminals with the buttons are scattered around a store, and customers are asked to press a button to show how they are feeling at the moment. That is all a customer is asked to do.That is all the information that the HappyOrNot company collects.
They collect an enormous amount of data. The buttons have registered more impressions than Amazon or Yelp. But of how much value is the data when there are so many variables which might affect a customer’s mood at a moment in time? One way of using statistics would be to to try to either control all the variables but one or two—a controlled experiment. Here, the data doesn’t give answers—there’s no way it could. But it turns out that the data has great value in sending signals about where there are problems. Owens gives some examples, such as a store that had the least customer satisfaction at the times when sales were greatest—-and therefore the salespeople were the busiest.
This is an example of the usefulness of big data. The cheap collection of massive amounts of data gives hints for later investigations with different information.