Interesting statistics from the tarsnap backup service
daemonology.net
daemonology.net
As for other interesting statistics, one of the more important one is life cycle, how long on average does an account live. If you multiply your number of signups in a given month with the lifecycle you can figure out an upper limit to how far your business will grow organically (without additional promotional activities).
Have you seen the interest rates banks are paying these days? Interest on unearned revenue is non-zero, but not by far.
As for other interesting statistics, one of the more important one is life cycle, how long on average does an account live.
I can't give you any accurate statistics on that, for two reasons: First, the number of tarsnap accounts which have ever closed is too small to be statistically significant; and second, because tarsnap doesn't have any fixed monthly fees, people can leave their accounts inactive forever and never have them die.
PayPal's first business plan was to make money on the float. That didn't work, so they started charging (ever higher) fees. In the history of the world, I don't think anyone's ever made their money on the float.
2) Makes an excess return with or without float.
3) Made his initial money without using float.
4) Is a fairly unusual guy, which is why we've all heard of him.
A more useful comparison would be to Jeff Bezos: Even disregarding unearned income, tarsnap produces roughly 1.5 months of float due to AWS billing a credit card at the end of each month (0.5 months after usage, on average) which has a billing period ending 3 weeks later (0.75 months) and which has a long enough grace period that I can wait another week (0.25 months) before paying off the credit card bill.
This "negative working capital" model is one of the reasons Amazon could grow so quickly in the early years -- and makes it very unlikely that I will ever need to raise capital to fund tarsnap's growth.
It makes accounting more difficult, and unless you are careful not to spend it, you risk a 'run on the bank' if a large group of your customers demand refunds.
And unlike a bank, there is probably no insurance for the customers' "deposits". Also, if you need to raise your prices, do you allow people who already deposited money to have access to the old prices as they might expect?
But it sounds like Tarsnap has addressed these issues carefully, and the transparency is a great thing.
Most startups deal with this by putting a ginormous WE DO NOT OFFER REFUNDS FOR ANY CIRCUMSTANCES disclaimer on it. Personally I always hated that so I offer refunds for any circumstances. Since only about 2.6% or so of customers ask for them, I don't need much of a cash cushion at all.
Me too -- tarsnap close their accounts and get a refund whenever they like. Not many do; but in the rare case that someone decides to stop using tarsnap, I'd rather refund their money than feel bad about keeping it.
I have no clue. If I could figure out what data people were storing, tarsnap wouldn't be a secure backup service, would it?
(For that matter, I don't even know how many files people have archived, or how large they were prior to being compressed.)