It's an arbitrary number, but so will the real estimate; what I think the important exercise is: take the arbitrary "real" estimate and have them back it out to what multiple of their projected sales that's meant to be.

As the other commenter points out: it's very difficult to gauge how valuable a share in a private startup is going to be worth. I think it's a lot less hard to reason about how much revenue the company is going to make.

That doesn't directly solve the valuation question --- the multiple on revenue is going to vary wildly from 3x to 100x --- but at least you get an idea of what they think their business plan is; and, if they're hoping for a 50x valuation, or they think they're going to be doing $1bn top-line in 2 years, then at least you know they're being crazy optimistic.