Hindman, Matthew2025-09-182015-04Hindman, Matthew. “Stickier News What Newspapers Don’t Know about Web Traffic Has Hurt Them Badly – But There Is a Better Way.” Shorenstein Center Discussion Paper Series 2015,D-93. Cambridge, MA: Harvard University, April 2015.https://dash.harvard.edu/handle/1/42719511In early 2000, Google conducted one of its first online experiments. The result was a disaster. Google’s experiment split off several groups of users to receive 20, 25, or 30 results instead of the standard 10. When Google checked six weeks later, they found – to their shock – that traffic had plummeted. Users given 30 results were doing 20 percent fewer searches. Google eventually traced this drop to a surprising source. It took Google half a second longer to return more results: 0.4 seconds to return 10 results, but 0.9 seconds to return 30.1 The most important lesson of Google’s experiment concerns what is loosely termed stickiness. Stickiness is like a compounded Internet interest rate: it measures how likely users are to visit, and how often they go beyond the first click to the second or third. Sites with above-average stickiness grow their audience share over time, by definition; those with below-average stickiness shrink. Site speed is one of hundreds of site features that affect audience growth. Over a day or two this slight delay meant little. But as the weeks wore on, the difference of that extra half-second was compounded again and again. People visited Google less often – and when that smaller group did return, they were a bit less likely to come back next time. Small traffic losses snowballed.Working Paper