Sales Diversity

What it is

Daniel Fleder and Kartik Hosanagar set out the problem in Management Science in 2009. Using models and simulations, they showed that some common recommenders can push each person towards new products while pushing everyone towards the same ones, so that individual variety rises as the market concentrates. They concluded that design choices shape the outcome.

A decade later Dokyun Lee and Hosanagar tested it in the field. Their randomised experiment on a large North American retailer's website, reported in Information Systems Research in 2019, covered 1,138,238 users and 82,290 products. Recommendations of the "customers who bought this also bought" kind lowered overall sales diversity compared with no recommendations. Individual shoppers explored more, but similar shoppers explored the same items, and while niche products gained sales, popular ones gained more. According to Carnegie Mellon's summary, the bottom 80 per cent of products took 37 per cent of sales without a recommender and 27 per cent with one.

The other side comes from Netflix. In 2015 Carlos Gomez-Uribe and Neil Hunt reported that their recommender spread viewing across many more titles, roughly four times as many on their own measure, than an unpersonalised ranking would. Dietmar Jannach and Michael Jugovac's 2019 review of business value puts the typical sales lift from recommenders at one to five per cent, which is the size of the effect whose distribution is at issue here.

In effect

On a typical retail site, expect suggestions to send you, slightly more than you would otherwise go, to what everyone else is already buying. A subscription service that pays for its catalogue in advance has a reason to spread viewing out; a retailer selling stock does not. Looking past the first row of suggestions is the simplest way to see more of the range.

What it does not say

The field evidence for concentration is one large randomised experiment at one retailer, read here at abstract level, with the 37 and 27 per cent figures from a university news release. The Netflix claim cannot be checked, because the company withheld the values on its chart's axes for competitive reasons. Neither result shows that recommenders reduce what an individual shopper discovers; in the field experiment individuals explored more.

---