Damnit, Keith. I wanted to submit this article and promote it, but now I'm afraid you've missed the HN success window.
Anyhow, I just read this and it's honestly the best article about customer acquisition and retention I've ever read, I think. I wish I wrote it. ;)
One small detail: your effective cost of customer acquisition is eCAC = CAC(1-k) where k is your viral coefficient.
In the article Keith only carries the discount forward one generation, but each generation begets another so it winds up being a geometric series.
So if k = 0.5 the discount on your CAC is 50%, not 33%. One spot check for this is that if k = 1 then eCAC should be 0.
*: This assume you can (1) accurately measure your viral coefficient and (2) the viral coefficient is the same for all users, regardless of how they were acquired. In the real world neither of these is generally true.
Thanks for pointing this out, I've corrected this article. Yea I didn't do much in the way of promoting it on HN but thought I might as well put it out there.
2 comments
[ 2.9 ms ] story [ 13.2 ms ] threadAnyhow, I just read this and it's honestly the best article about customer acquisition and retention I've ever read, I think. I wish I wrote it. ;)
One small detail: your effective cost of customer acquisition is eCAC = CAC(1-k) where k is your viral coefficient.
In the article Keith only carries the discount forward one generation, but each generation begets another so it winds up being a geometric series.
So if k = 0.5 the discount on your CAC is 50%, not 33%. One spot check for this is that if k = 1 then eCAC should be 0.
*: This assume you can (1) accurately measure your viral coefficient and (2) the viral coefficient is the same for all users, regardless of how they were acquired. In the real world neither of these is generally true.