I'm somewhat puzzled by this paragraph - even moreso as they seem to base a lot of advice on top of it:
For Facebook, they’ve defined core retention behavior as the point when a user has added at least 15 friends to their account. In Alex Schultz’s Startup Class lecture he discloses that, for Facebook, someone that has 15 friends is likely to remain active on the service for a long time, so their retention strategies are all focused around promoting the behavior of adding friends.
To find your own core retention behavior, look for specific actions that are common to your ideal users, then promote that behavior in your retention strategies.
I'm not a marketing expert, but this seems like a classic example of "confusing causation and correlation" to me. So, in the example above, facebook has found that "user has > 15 friends" is a good indicator for "user is loyal". Fair enough. but instead of going "we have this awesome way of identifying loyal users, now let's find out what made them loyal in the first place", the guide assumes that the high friend count actually causes the loyality and advises to do everything to bump it up.
In Facebook's case, this probably works, because friend count might actually influence retention in some ways. (e.g., your news feed might become more active or more interesting, making you want to stay). But even then, you're kept completely in the dark what made those people sign up and add friends in the first place.
i think taking this as a general advice might actually be dangerous. In the best case, you keep a high retention rate but have no idea why; In the worst case, it might tempt you to build "features" that try to manipulate or even force the users into particular behaviors, only to inflate some metric. (see, e.g. Pinterest's "auto-following" routine: http://www.businessinsider.com/wait-a-minute-pinterests-sign... )
Comments
I'm somewhat puzzled by this paragraph - even moreso as they seem to base a lot of advice on top of it:
For Facebook, they’ve defined core retention behavior as the point when a user has added at least 15 friends to their account. In Alex Schultz’s Startup Class lecture he discloses that, for Facebook, someone that has 15 friends is likely to remain active on the service for a long time, so their retention strategies are all focused around promoting the behavior of adding friends.
To find your own core retention behavior, look for specific actions that are common to your ideal users, then promote that behavior in your retention strategies.
I'm not a marketing expert, but this seems like a classic example of "confusing causation and correlation" to me. So, in the example above, facebook has found that "user has > 15 friends" is a good indicator for "user is loyal". Fair enough. but instead of going "we have this awesome way of identifying loyal users, now let's find out what made them loyal in the first place", the guide assumes that the high friend count actually causes the loyality and advises to do everything to bump it up.
In Facebook's case, this probably works, because friend count might actually influence retention in some ways. (e.g., your news feed might become more active or more interesting, making you want to stay). But even then, you're kept completely in the dark what made those people sign up and add friends in the first place.
i think taking this as a general advice might actually be dangerous. In the best case, you keep a high retention rate but have no idea why; In the worst case, it might tempt you to build "features" that try to manipulate or even force the users into particular behaviors, only to inflate some metric. (see, e.g. Pinterest's "auto-following" routine: http://www.businessinsider.com/wait-a-minute-pinterests-sign... )