Best Model to Predict Referrals (Viralatity / Network effects)

[scruz9]

All,

I'm looking to put a model in place to predict the number of new users from my existing customer base through referrals. It looks like an expotential curve fits the data best, but I'm looking for other models that might be a better fit. I plugged in sample data from a preexisting cohort Columns A, B to calculate future periods. I've been recommended to use a lognormla distribution, but not too sure how to put that into place in excel. Any help would be appreciated.

[shg]

The attached workbook uses Solver to calculate a best-fit lognormal distribution.

[scruz9]

Hmm looks like a LogNormal Regression wouldn't be the best solution for this model. It seems to die off too early and be too aggresive around it's peak. Well it was worth a try. Unless I see any better fitting models, I'll use this expotential decay curve for future values.