I'd go farther, if players stick around 12-15 months historically, and they had 3M users, then in 9 months they could lose 60%-75% so 1.8M to 2.25M players! This of course depends on distribution of lifespans, and assumes that the 12-15 month figure remains accurate.
If that's true, they're getting users for as cheap as $50 each, and making $100 on each.
I would have spent at least part of those $120M in funding startups to develop new games (and later bring them in, in case...). I'm pretty positive the return might have been better...
Losing more users is good in that it means they gained more in the same timespan - for fixed changes in the number of users (which is the datapoint we have). But it also means they are having a hard time keeping them.
This is an obvius mistake, as not all of the 3 Million from Dec 2010 continued to be a customer in Sept 2011. I roughly assume they lost 800.000 customers in that span (i think its more), then the newly aquired customers triple to 1.2 Million. Thus Zynga makes 50$ profit per customer.