Having briefly followed Clay Shirky’s critique of Second Life over the holidays, I began thinking about what we can learn from SL. In a nutshell, he says SL is overhyped because of all the emphasis on the registration numbers hides what’s really going on (less than you would think) and that the previous failures of 3D worlds (VRML, etc.) prove that other online communities are more viable and therefore, more interesting. Even Danah Boyd is getting testy over the Second Life hype, primarily because people are talking about “living virtually” and she believes technology should enhance people’s existing sociability, not replace it. All are valid criticisms in my mind, but not necessarily lethal ones. However, it did spark me to think, what can we learn from Second Life that we can apply to other things, in general?
SL has two interesting charactistics: 1) SL is a community; until you start participating with other people, you haven’t really experienced it to its fullest, and 2) Linden Lab does not spend money on traditional advertising, so much of the growth can be attributed to community marketing and PR. I thought I’d take a look at traffic numbers and see if they tell us anything about communities. This post is the first in a short series on what I think we can learn about the traffic numbers of various sites and comparing them.
I’m working on a hypothesis, but these are just observations right now. What I’m hoping to discover is a traffic curve that’s a tell-tail sign of a vibrant community; as you’ll see below, Myspace.com may be that prototypical traffic curve.
All the following graphs come from Alexa.com. I set the resolution so that I could view the traffic numbers for the life of the website, or almost all of it. This eliminated websites that existed prior to 2001 because Alexa only reports on data within the last five years.
First, the traffic graph of Secondlife.com, which I’m using as a proxy to the SL client. 
It looks like a slow build up for the first few years, corresponding to the “good ole days” when there were a relatively few, but active residents. Then, around mid-2005, interest begins to twitch to life. By 2006, there’s a growth curve, with spikes that are most likely generated by stories in the media.
If you were to fit a curve to the data, it would look like a long-tail at the beginning, with growth towards the end. The dip at the end could be attributable to either 1) hype no longer working, or 2) the holidays. Time will give us more data to tell.
Let’s compare this traffic curve to a few other community traffic curves. First, myspace.com.
The myspace.com’s traffic curve looks much more like a classic adoption curve. You can see where the Second Life curve looks similar to the 2004-2005 portion of the myspace curve. However, this looks more like a mature adoption curve, consisting of three sections:
1) Small, but active community bubbling along for a year or so,
2) Followed by a period of growth–which oddly, doesn’t look like it’s affected that much by the spurt in Q4 ‘04, which was probably fueled by media stories.
3) Followed by a slowdown in growth beginning after Q1 ‘06.
Let’s look at another community site, Drupal.org. Drupal is the popular open source content management system that Howard Dean’s DeanSpace was built on top of. It has a similar shape to Secondlife.com’s traffic curve: 1) a long tail at the beginning followed by 2) a growth curve starting around 2005 and gaining traction in 2006.
Finally, Wordpress.org, the website for the blogging software that I use for this blog. Again, a similar shaped curve.
Looking at successful open source software projects is appropriate because they are only successful if there is a vibrant community around them.
Note that all the curves show a decline at the end of the year, lending credence that the decline in Secondlife.com is due to the holidays rather than a natural decline. Also note that none of these curves have the same horizontal-ish section on the right that myspace has, possibly meaning they have yet to reach the mature section in their life. If myspace has indeed matured, then it becomes a cash-cow, but, but may be impossible to grow larger.
Coming up, I’ll look at traffic curves for other popular marketing websites. Then, I’ll follow that up with superstars. Hopefully, I can develop a testable hypothesis on what constitutes a viable community website with some idea of roadmap people can use to determine what stage their community is in. Or, I could fail miserably. Either way, I’d love to hear your thoughts.
8 responses so far ↓
1 Idea City - New Thinking about Marketing, Advertising, and Culture » Contrasting Community Sites with Marketing Sites // Jan 4, 2007 at 3:31 am
[…] One Metric that Describes a Community? […]
2 Clay Shirky // Jan 4, 2007 at 9:25 am
First, the traffic graph of Secondlife.com, which I’m using as a proxy to the SL client.
Why are you using the traffic graph as a proxy? It is unrelated to use. Traffic to the website only measures interest, not adoption. A lot of people sent away for Sea Monkeys too, because they liked what they saw advertised, but that said nothing about how many were satisfied by their purchase.
Crucially, no web numbers measure anything about SL use — the only source of real SL numbers is Linden, and they are obviously not in the mood to report anything about actual use. Making assumptions about SL using SL.com is like that old joke about the physicist betting on the ponies: “First, I assumed a spherical horse…”
3 Fulminator » Blog Archive » Measuring Community Value // Jan 4, 2007 at 7:31 pm
[…] However, there are some interesting pieces here and no matter the limitations in the data used to derive the theories, they are interesting topics to think about. Greenberg basically compares site traffic over time to marketing sites and community sites and looks at the patterns. It started with a post about Second Life (and my chart reading is much more negative regarding it than his — not sure it’s just the holidays) that didn’t convice me. Then in his follow up post he starts to lay out more interesting examples. On one level, this isn’t much different then television exposure and wearout analysis. But it’s interesting when applied to the web. […]
4 Joel Greenberg // Jan 5, 2007 at 1:41 am
Hi Clay,
Thanks for your thoughts. My primary goal for writing about this is because I work in the marketing world. I thought I’d be able to contribute to the discussion of the value of community. Many people are trying to put a value on community so that they can convince advertisers to buy into community rather than put money into, say, direct mail, TV, etc. If you had a community site and you were looking for ways to make money so you could continue to serve your community, and if you saw advertising and marketing as one way to make money, then the question of putting value on your community is critically important. That’s the callenge.
I’m using traffic as proxy for SL community for a couple of reasons.
1) It’s easy data to get via Alexa, so at least I have something to chew on other than opinions–despite Alexa’s limitations. If I were to rely on actual traffic numbers on all the different sites I cited by asking them, we wouldn’t be having this discussion. We’d be waiting to hear back from them…maybe forever.
2) Secondlife.com is in fact a community itself. Certainly the same, in kind, as boingboing.net–a blog with active readers and commentors. Up until this summer, it had an active bulletin board system, so historically it behaved as many other non-blog communities did. The people involved in this online community are also involved in-world, so I think it’s plausible to assume that the activity on Secondlife.com could somehow be related to the activity in-world, at least for this thought experiment. As an aside, I think it’s interesting to note that there is alot of community activity related to Second Life on the web and not occuring in-world. No one’s really talking about that yet, but I think it’s an interesting avenue for exploration.
If I could get the in-world login data from LL and their traffic data from their log files (or the traffic data from Alexa),we could run a simple correlation to see how much traffic to Secondlife.com is correlated to traffic in-world. Currently, there’s no easy way to get that data, but I still think there’s value in having these types of discussions. Yes, website traffic hides both interest and adoption into one metric and obscures both, but we’ve got to start somewhere.
It’s important to clarify that I think what’s most important is the shape of the curves, not the absolute traffic numbers. Clearly, people are participating in Second Life, so it is a community and maybe studying particular metrics will allow us to recognize communities better. If I’m on to something, then it’s easier for me to go to a client and say, “What would you rather do, put your marketing money into a quick viral hit that fades, or into a community that stays? Here, let me show you what I mean…” Doing it right, the community benefits as does the marketer.
There is something interesting going on. I think Second Life is qualitatively different than the VR says of the early ’90’s and the VRML days of the late ’90’s. Aside from the advanced in technology, the big difference between now and then is people. Sure, the community isn’t as big as the registration numbers would lead a casual observer to believe, but strip away the hype and there is still something interesting going on. As of Dec. 29th, there were 534,738 logged in-world in the previous 30 days (sourc: http://secondlife.com/whatis/economy_stats.php).
So, let me ask you, having studied communities all these years:
1) What would the metrics be to measure community? In other words, how would someone from outside the community be able to make a rational decision on whether or not they want to advertise to, market to, or engage with the community on a business level and then compare that amount and effort with other ways of reaching people?
2) Do you see similar traffic patterns with other viable on-line communities? (Look, I’m not hung up on traffic, but it’s a start.)
3) What are common engagement metrics for community? (This is important in marketing because there’s no commonly agreed upon measure of engagement, only reach and frequency, yet many believe engagement has to matter.)
I believe LL would be willing to share data. I’m sure if we were to approach them with a request, they would share the data with us. I’m thinking a consortium of business and academia would be appropriate. Are you interested in discussing this idea further?
5 Idea City - New Thinking about Marketing, Advertising, and Culture » Community Superstars? // Jan 5, 2007 at 6:43 am
[…] One Metric that Describes a Community? […]
6 Putney Swope // Jan 12, 2007 at 6:43 am
How is it that Greenberg can objectively study a community in which he (or at least his company) has a vested intrest? GSDM has spent a great deal of money to have a presence in Second Life; money they hope to recoup (and then some) by charging their cleints for adding Second Life in some form to their marketing programs. And what is more attractive to an advertiser: an audience of 2.5 million or and audience of 250,000? Th former, of course, which means it is in GSDM’s (and hence, Greenberg’s) best interests to make Second Life seem as popular as possible.
7 Joel Greenberg // Jan 13, 2007 at 1:14 am
Putney,
I appreciate your skepticism and understand why on the surface, bigger appears better. However, it does no service to our clients to inflate numbers. It may convince them to do the first program in Second Life, but would work against them pursuing further programs. I believe there’s long term value in Second Life, not quick hits. Honesty is the best economic policy.
I think a more useful number would be the ratio of residents that have signed in within the last 30 days to total residents, which is currently 21%, and to map that over the past three years. That would show the growth.
My hunch is that the ratio would be fairly steady. (I got the numbers here:http://secondlife.com/whatis/economy_stats.php). ) The problem with the Second Life stats is that they are a snapshot, and not historical. I’m sure Linden Lab would provide us data if we asked nicely.
So, Putney, what numbers would you like to see? I think it’d be great for us as a community to reach a consensus on what numbers we’d like to see and then go to LL. I’m sure they’d be happy to give us the data. I don’t think they would be able to accommodate lots of going for one-offs.
Joel
8 sonia // Mar 11, 2007 at 1:23 am
I am also interested in the historical total traffic data of SL in-world. Unfortunately, I’ve requested them several times from LL, but LL’s response is far from encouraging (see example below).
LL: This is a feature, not a Knowledge Base request. You can propose a new feature @ http://jira.secondlife.com , read the instructions and get fellow Residents to vote. Note that high votes don’t guarantee implementation…
If Joel is able to get the numbers from LL, please share with us. I believe a heavily researched area can grow faster.
Leave a Comment