Update, 10/8: The story continues here.
Earlier today, the great Virtual World News blog (run by the folks putting on this month's San Jose conference that this blog is a media partner with) ran a press release from Yankee Group, a Boston-based analyst/consulting firm, summarizing their latest report on Second Life, and it includes this strange passage:
"[T]he growth rate of Second Life users has slowed since its peak in October 2006, while user engagement (as measured by average time spent per user) has leveled off at just 12 minutes per month." [emph. mine]
I've looked at that sentence several times, and can't make head or tail of it, because it's totally at odds with all meaningful available metrics. At the end of October 2006, there were about 1 million total Resident accounts; today, it's about 9.8 million. In December 2006, maximum concurrency of in-world Residents was peaking at about 20,000; in recent weeks, it's been maxing at around 50,000. (Last Sunday, according to Tateru, it reached just shy of 52,000.) In terms of active monthly recurring users, in October 2006 it was about 150,000; last couple months, it's been about 550,000. The "twelve minutes per month" claim is the strangest, because according to the Lindens' published demographics, on average, active monthly users are in-world between 42 to 58 hours per month. (See inset chart.)
The Yankee Group's press release offers no indication of methodology (hence this post's clever ass title) but near as I can tell, they're counting usage activity of all 9.8 million Residents. Which as longtime readers know, is a dubious proposition, since only 70% of that number represents unique users (and not alts), and just 10-12% of account creators return to Second Life after a single visit.
Does Yankee count one-try visitors as "users"? I'm guessing so, because then the 12 minutes claim would probably make more sense then-- as more and more people try Second Life, more and more find it too complicated/frustrating/apparently purposeless, and leave after a few hours. If you count their in-world time anyway, that would drive down the total "user engagement" minutes. But then, that really wouldn't capture actual activity of regular users, either-- only measure the majority who bounce off the system.
Maybe I'm missing something. For example, maybe there's a typo, and the Yankee Group is actually writing about an entirely different world I'm not familiar with. In any event, I sent off an e-mail to the firm's listed analysts, and will publish any feedback they send in an update here.
Update, 10/4: Report co-author Christopher Collins e-mailed me a couple days ago to say, "All of my stats came from my analysis of published Linden Lab data." I've invited him to explain his methodology in this post's Comments and will run an update if he does so.