In my "Sherpa" work with startups, one of the things I wrestle with (especially in "bubble" periods, like the present) is what metrics are best to show VC's when raising money. The issue is a complex one, and easy to oversimplify, but I do think that there's merit in the advice that it is better to have rapid growth in some user metric (user growth or engagement being the primary ones) than it is to have revenue, at least small amounts of revenue (regardless of growth rate).
From experience, I often think the best advice is that it's better to have either lots of growth in users, or lots of revenue, but NOT a small amount of revenue (even with good user growth). Once you have even $1 of revenue, you invite VC's to employ "traditional" valuation procedures (e.g., some line item of your income statement times some "industry multiple") that, given the small multiplicand, will not generate a large valuation.
In bubbles, rapid user growth/engagement (without the messy matter of demonstrated monetization) allows the VC to believe that trees will grow to the sky. This "triumph of hope over experience" (which seems to happen about once a decade) is exacerbated by the competition among VC's for the next hot deal (which competition is further fueled by the pressure from their limited partners). Sadly, this pressure is based on some actual statistical evidence that the VC firms that do the best are those who find the "hot" deal (and, very importantly, who exit to some safer asset before the bubble bursts).
The required suspension of disbelief is easier when there are no hard metrics on monetization to analyze -- the bubble of the dream of infinite user growth is kept aloft by the belief that, somehow, either (a) one will be able to figure out how to eventually monetize the infinite user base, or (b) sell the company to a greater fool who believes that they can.
An interesting note on this by Nick Bilton in today's NYT. Which has evoked a lot of debate in the start-up echo chamber (including, of course, this post).