Here’s a (perhaps weak) alternative explanation for the headline and chart of the graph: suppose there is some UX quality number representing how good an average app will be if it is updated. Now suppose this number increases at 10% a year. If you update your ux every year, you might therefore observe a 10% improvement each time you update. If you update your app only every 5 years, you would instead see a 60% improvement.

So my question is: how does this study compensate for changes to the frequency of updates?