Nudging your way to five-stars
Consider this quirky data point from Uber: only 1% of trips in San Francisco gets a one-star or one-star rating according to their 2014 blog article. Even after accounting for the systemic screening of drivers and the fact that Uber drivers may not be a random sample of drivers, the 1% figure ‘feels’ too low for a population of at least 20,000. What could explain such a skew?
A few days ago, the driver (or driver-partner as Uber would like to call him) of an Uber cab I hailed belched through out the ride – disgusting, right! I had made up my mind to give him a one-star. At the end of the trip though, he did a curious thing – he rated me five-star and showed it to me, saying, “Sir, I gave you a 5”. This blog post is about the consequences of that act and how it goes a long way in explaining the skew.
Uber India drivers are master behavioral scientists, as are gold merchants and sari retailers (but that’s a subject for a different blog). The cleverness lies in their use of reciprocity to nudge an upward revision of the rating. Drivers, when ending a trip, are shown the rating screen – this happens usually when the rider is still in the car. To be a truly fair system, one would expect the driver to conceal their rating of the rider and vice-versa. Surprisingly, Indian Uber drivers rate the riders five-star and cheerfully show it to the rider. This simple act kicks in reciprocity in the rider. Suddenly, the rider is under pressure to revise the rating he had ‘decided’; northwards. From ‘this belching buffoon needs a rap on the knuckle and a one-star’ to ‘maybe I should give him a three-star, after all he gave me a 5’. Robert Cialdini, in the book ‘Influence’, says “we are obligated to the future repayment of favors, gifts, invitations, and the like” – he would be proud!
How would we design a system that doesn’t lend itself to being gamed? For a start, Uber could add a forced delay between the end of a trip and the feedback step. A one minute interval could be enough to retain the memory of the last trip (to ensure accurate feedback), while forcing distance between the rider the driver. It’s not a fool-proof system – drivers could still orally indicate the fact that they would be giving a five-star rating. But I wager that the ratings would be far more evenly spread than the 1% one-star skew Uber has presently – one that is representative of the ride!
Image Credit: http://findingthefreedom.com/using-uber-abroad/