On Plurk I noticed Hamlet’s article: New SL User Retention is 20% (Wha?) I mentioned that stat in one of my ‘Rod Interview’ articles. Rod had said of the 400,000 new sign ups to SL each month 20% stay longer than 30 days.

Stats – Do they Work?

Hamlet was thinking about that and wondering about the numbers. Hamlet had thought about past numbers and remembered that previous monthly sign ins was in the range of 500,000+. If there are 1,000,000 sign ins per month now and 400,000 of them are new sign ups each month, Then that means SL has only grown to 600,000 over a number of years. How can 20% stay and not drive up the sign in count? That should add 80,000 users per month.

It is a good question. Obviously as many have to be leaving as are coming in for the numbers to remain the same. The equation is simple addition and subtraction.

Regular Sign-ins + New Sign-ins = 1,000,000 – Since new sign-ins = 400,000 Regular Sign-ins must be 600,000. Since 20% of 400,000 is 80,000 one would think we should add 80k to the 600k for next month. That would give us 680,000 for the next month and 760,000 the month after and so on. But, that is not happening. So, some steady stream of people has to be leaving SL for the number to basically hold steady.

I looked back at Metaverse Building’s study of user longevity in SL. (here) At the time they were looking for the 2-year Effect. But they found 3 longevity groups. The 1 to 400 day avatars, 400 to 1300 days and 1300 and longer.

About 66% of users considered fell in the 400 day group, dropped out within 400 days.  Only 0.5% had stayed 1,300 days (about 4 years). About 5% last 2 years. There isn’t really a 2-year Effect in SL.

The trend in people leaving is shown by the data to be exponential. So, the 80% drop rate in the first 30 days is likely right as it seems to more or less fit with Metaverse’s data. They see 32% dropping out within 100 days and 48% in 200 days. Presumably the majority of the 100 day drops outs left in the first 30 days.

The three groups of people in SL explain why the user sign-in count is growing slowly.

How accurate this is depends on how much the current SL is like the 2007 to 2009 SL that Metaverse looked at. I expect the numbers have shifted, but I think the trend shapes are likely the same. So, we could estimate the numbers from what we have. But, they would be weak estimates.

Suffice to say, SL has amazing durability and is doing surprisingly well for a 10 year old ‘game.’