It’s time for some more pretty pictures to help answer a debate between myself and fellow OSGrid admin Nebadon. The core question is essentially; “Have the users who joined us since the LL OpenSpace price change, stuck around?”. The methodology we are using is based on last login times – it’s the only bit of data we store relating to how active users are, that we can query in a reasonable period of time.

The three questions we will ask of the database are; “How many unique users logged in in the last day? week? month?” – of the 10,958 who have at least one recorded login in the last 6 months since we added the field, 468 logged in today, 1,142 this week and 2,822 this month. The next question to ask is – how does this break down across by signup date? The following highly compressed chart shows the information for the number of logins in the last 24 hrs, 7 days and 30 days from each signup group. (A signup group is all the users who signed up in a certain calendar week).

Fig 1. Logins versus Signups

You can still pretty distinctly see the point at which Linden announced OpenSpace price changes (Y2008W41), but it doesn’t quite explain if these users are sticking around more, less or the same as users who signed up beforehand. We’re expecting some kind of drop-off since a lot of users were just experimenting with alternatives – and we’re not really focusing on being a SL clone (unlike some of the other grids). So, the next chart shows the above data – but divided by the total number of signups in that week.

Fig 2. Signups versus Logins Proportionally

I clipped off some of the earlier data since it was giving erroneous values, but what is interesting here is that older users have a fairly incredible retention – >30% of users who signed up in 2007 / early 2008 are still around and logging in. Newer users have dropped off – but there’s still a pretty high proportion who stayed – both myself and nebadon believe these might be region operators – early users were all region operators, and region operators probably have a pretty good retention.

So let’s take a look at some more data – the following is logins from the last 30 days – how many are region owners (cross referenced against signup date)

Fig 3. Region Owner versus Logins

Now this is a little incomplete, since it wont measure things like estate managers – it only shows people who are the registered region owner by the Master Avatar parameter. There is a trend for people with older accounts who still login regularly to be more likely to own a region – but it’s not absolute.  Looking at this data you might be mistaken for thinking that newer users are less likely to own a region – but charting the data % of region owners by signup date you get a different picture.

Fig 4. Proportion of Region Owners

In conclusion – the data seems to indicate mixed results to the original question; users who signed up looking for an SL alternative are more likely to have left than users who joined beforehand – but trends seem to be indicating the core underlying growth is still pretty strong; and a healthy percentage of the people who did sign up in that period have stuck around – overall growth of active users increased by between 200 and 600 percent after that jump.

There is also a trend that the more active users are region owners – of the people who logged in in the last 24 hours (versus 30 days for Fig 3), closer to 50% were region operators. This doesnt strike me as too suprising given that OSGrid’s real strengths are in the ability to connect in regions free of charge – more active users are more likely to b involved in that aspect of the grid.

For other grid operators who are interested in calculating the same numbers I have above for comparison, on the next page are the horribly complex SQL queries I have been utilizing to generate this data. They should work with any OpenSim-standard MySQL installation. Click here to see them.