3 ms·
I was curious: when optimizing a leading indicator, what sort of effect does that have on said leading indicator? Presumably there can be a correlation-not-cau
by sbov 14y ago
I was curious: when optimizing a leading indicator, what sort of effect does that have on said leading indicator?
Presumably there can be a correlation-not-causation effect, and optimizing said indicator wouldn't have as much as an effect on engaged users as it did on the indicator itself. Does anyone know of a situation where they had an increase in the leading indicator but none in engaged users? Might the relationship change over time?
- RichardPrice 14y agoIt's a good question. You need to find a leading indicator that really is causal of engagement, and not just tangentially correlated with it. Chamath stressed how critical it is to both minimize the time it takes to deliver a user their first 'Aha' moment, and then to continue to deliver those moments as regularly as possible. Presumably how many friends you have is causal, to some degree, of those 'Aha' moments - I expect that many of the 'Aha' moments on FB are to do with seeing cool content from your friends. Nabeel also mentioned that it's also important to have various KPI metrics that you are aware of. His view was that there should be one over-arching operational metric that is broadly predictive of future engagement, and that the growth team should focus on that, but that, for any given tactic that you try out, you should be aware of its effect on a number of KPIs.