Thank you for this very useful model. I was investigating the subject a bit and found an edit of the Chen model that accounts for variable “churn” rate (http://bit.ly/9hjaPE).

It would most certainly be interesting if this feature could somehow be implemented in your next iteration. The author of the modified Chen doesn't appear to justify the variable churn rate, but it does sound logical in many ways: a) as you gain loyal customers over time, they may send more meaningful invites to a better targeted audience, b) as you learn from experience and tweak your content or service you lower attrition, c) as customers find more of their friends using a service, they are more reluctant to leave. There are probably a dozen more explanations that look good on paper, but would they make sense in the Excel spreadsheet?

I don't want to say Survival Analysis, but that immediately springs into the [very] back of my mind from college stats class.

Thank you for the model and I'll be looking forward for your next version!

If carrying out the model over a period of three years, is it wrong to do the projections for year 1 and use the figure generated at week 52 as the seed amount for year 2?

It can be a bit of a mind twister indeed, but the retention rate is independent of the cycle time. The coefficient and cycle time determine how many and how fast new users come on board. Once on board, the retention rate determines how long you keep them.

In many companies I have been involved with, the two jobs are in fact carried out by independent teams.

I have just started my career in Marketing and am being introduced to these concepts of social media and viral growth. I find it all fascinating and your viral growth model has been of exceptional help. I do have a questions re: the TAM and Retention Model, though. In the preceding example, you used a cycle time of 1 week and converted the annual retention rate to a weekly attrition rate. If the cycle time were changed to 2 weeks, would you then convert the annual retention rate to a bi-weekly attrition rate, using the formula (1 -(Annual Retention Rate^(1/26)) rather than (1 -(Annual Retention Rate^(1/52))?

Thanks so much!