Yes, I'd love to see a more robust approach to novelty decay rates (to make up a metric) in A/B testing, especially around engagement.
You'd think that would be of interest to the business to know that X "successful" intervention had a typical average lifetime of Y before reverting to the mean, for example.
Then again, there's the idea that competitive advantage for social networks is just finding enough novel interventions before the novelty of the platform itself is exhausted, and FB has no short supply of other platforms to milk in that regard.
You'd think that would be of interest to the business to know that X "successful" intervention had a typical average lifetime of Y before reverting to the mean, for example.
Then again, there's the idea that competitive advantage for social networks is just finding enough novel interventions before the novelty of the platform itself is exhausted, and FB has no short supply of other platforms to milk in that regard.