Second Life posts weekly uniques

 Posted by (Visited 9194 times)  Game talk
Oct 032006
 

I’ve mentioned before that weekly uniques is my preferred metric for measuring virtual world populations. You need to have an SL account to see the stats, but the announcement is here.

Drum roll, here they are:

Residents Logged-In During Last 7 Days 112,918
Residents Logged-In During Last 14 Days 150,554
Residents Logged-In During Last 30 Days 235,311
Residents Logged-In During Last 60 Days 351,206

Now, interpreting these by themselves is hard. First off, the figure for the week is quite healthy. The way that the numbers rise so hugely over the course of two months indicates one of two things:

  • lots of folks bouncing off the game, or
  • lots of folks with very sporadic login patterns.

The long-standing MMORPG pattern we’ve seen is that most days, half the active userbase logs in. Over a week, it’s a significant percentage of the subscriber base, and in a month, it’s everyone who isn’t quitting, with the exception of those who forget to cancel.

This Second Life pattern looks quite different. Anecdotally, what I have heard is that both the above are true — lots of folks bounce off, and also lots of folks pop in just occasionally, rather than making Second Life a daily stop.

The stats posted also include more economic data. BTW, for those interested in PCU, there’s this page.

  27 Responses to “Second Life posts weekly uniques”

  1. to get, but I’m simply not cool enough to be in their elite squad of commentators. It’s probably just easier to assume you’re wrong, whatever it is you meant to say, but since I have no idea what you said, there’s nothing to argue. From here

  2. [IMG] Cha-Ching http://www.chachingapp.com/ [IMG] Raph%u2019s Website � Second Life posts weekly uniques https://www.raphkoster.com/2006/10/03/second-life-posts-weekly-uniques/ [IMG] PhillRyu.com – Cha-Ching: The First Nail in Quicken’s Coffin? http://phillryu.com/2006/10/04/cha-ching-the-first-nail-in-quickens-coffin/#comments [IMG] Tuning and Optimizing RHEL for Oracle 9i and 10g Databases (Red Hat Enterprise Linux,

  3. Raph’s Website – Second Life posts weekly uniques

  4. You don’t judge a country by how many people are born and die each day, or how many come and go. That’s a MMORPG hangover.

    You just a country by its GNP, and you judge its economic health with things like “starter homes” or “home purchases” etc.

    Here are numbers you need to look at to understand the world’s stability, and level of civilization:

    o number of landowners in June 2005: 6,000
    o number of landowners in October 2006: 19,000

    While the numbers of landowners have nearly quadrupled, the percentage is still very small compared to the total number of log-ons. That’s because a lot of people rent — but even adding in the renters, you will still have a relatively small percentage.

  5. Judging a country solely by landowners is kinda one dimensional too. Knowing how many people are residents, workers, tourists, and so on is a valuable piece of data. For example, you yourself use it to arrive at your ratio of landowners to total visitors. 🙂

  6. Raph-

    I ran a quick and dirty little analysis based on that sites stats.

    Couple things to note:
    1. I lost some data (approx 5 days worth) due to the sites formatting.
    2. I didnt run any analysis of dates mainly related to the fact I didnt want to wrestle with English date structures in excel at this late hour.

    Anyhow this give you a breakdown of population as a percent of total. And total expendatures as a % of population type, based on historical data. The raw data, and the raw historical analysis are included. I’ll let you interpret as necessary, interesting to say the least….

    PS: Please excuse by rather bland and not-quite-finished blog site, have not really had the time required to make it pretty, start-up’s ya know 🙂

  7. THere’s some sort of spike in there for registrations. Wonder what it is. I didn’t bother looking at what date it was.

    The peak users data shows a classic daily and weekly cycle, with a general rising trend.

    I’m not sure what some of the data columns in your spreadsheet mean exactly, though. Also, how are “active users” defined in the original data?

  8. If you assume that every committed player (C) logs in once a week, and that every browser (B) only stays for a week, then you get the following equations:

    351000 = C + (60/7) B
    113000 = C + B

    Solving both of these equations, you get (approx): C = 83,000, B = 30,000. IE: 83,000 committed SL players, with 30,000 browsers per WEEK.

    Changing the period from 7 to 14 days, you get (approx) F = 89,000, B = 61,000 per 2 WEEKS. (or 30.5K/week, similar to above)

    If you change the period from 7 days to 1 month, you get (approx) F = 119,000, and B = 116,000 per MONTH. (or 29K/wk, similar to above)

    Which means that somewhere between 80K and 119K people are active participants in SL. 1/4 to 1/6th of them are land owners (aka: directly paying customers).

    The peak current usage is around 10K. The daily low is probably 5K (given other MMORPGs/MUDs). The average is 7.5K. 7500 x 24 hrs/day x 7 days/wk = 1.26M hours played/wk. Divide by 80K or 119K => average SL player is on someplace between 15 and 10 hrs/week. (This calculation excludes browsers, assuming that most of them are only on for a few hours and then they leave.)

    30K browsers/wk * 52 wks = 1.5 million people passing through per year.

    => SL needs 1.5 million browsers to come through per year in order to maintain a player base of 80-90K, and 20K paying players. (In Shareware metrics, this is a 1.2% conversion rate, which is normal. However, SL gets at least $72 per “conversion” ($6/month), which is much more than most shareware programs get.)

    Disclaimer: I may have made math errors and bad assumptions.

  9. You just [sic] a country by its GNP

    So what’s its GNP?

    You don’t judge a country by how many people are born and die each day, or how many come and go.

    2.5 hours ago, I got up and stepped out of an alumni lecture on fairness and factions in global health. Don’t tell me you can’t judge a country by how many live or die. Those rates mean a lot, not by themselves, but no MMORPG judges them by themselves, either.

    THere’s some sort of spike in there for registrations.

    September 11th (16K jump) and September 13th (22K jump). All the other jumps were 4 digits, at most shaving 9000ish. I scanned manually.

    I’m guessing SL did a Sept 11th promo and hit well. *glances over at TN* That should’ve hit TN. Now I’m curious, too.

  10. Theres a redirect now on that link you provided initially.

    If I recall from the data set:

    Total users were total registered users
    Total Actives were total active users
    Residents Online were current residents online
    Amount Spent was Amount spent in last 24 hours

    This was all stratified by:
    Date
    Time Stamp (Hourly)

    Briefly:

    So the data gives you a rolling number:
    Registered Users
    Active Users
    Residents Online
    Amount Spent

    Thus in looking at the historical stats:
    Minimum was the lowest number in all categories for the data set
    Maximum was that maximum in all categories for the data set
    Same with median/mean and standard diviation.

    Column D Gives you:
    Active users as a % of total residents by Min/Max, Median/Mean

    Column F & G Gives you:
    Residents online as a % of Total (F) and Active (G) by Min/Max, Median/Mean

    Column I, J and K Gives you:
    USD Spent as a % of Total, Active and Online by Min/Max, Median and Mean

    This data is a rolling number by hour, and date.

    So basically what were seeing here is:
    (if we take the mean in relation to the Maximum/Minimum amount)
    Are the number of active residence rising as a percentage of total? (Col D)
    Answer: Only slightly
    Are the residence online rising as a percentage of total? (Col F)
    Answer: Only Slightly
    Are the residence online rising as a percentage of active? (Col G)
    Answer: Definately
    (Thier population of active residence is online more!)

    USD:
    Same as above, however remember the amount spent as a % of total population, active poulation and online population is a rolling 24 hour total.

    Meaning they’re making a mean of 50.12 per online resident, 1.23 per active resident and .54 cents per person in the total resident population every 24 hours

  11. That should’ve hit TN.

    Actually, I can think of several things lately that should have hit TN…. the gambling stuff is another example.

  12. eh anyhow the analysis couldve been better but my time tonight is limited and it appears the data sets gone missing or been locked down on that linky. I’ll stop back by tomorrow and clarify the columns further if required, anyhow hope it helps…

  13. […] Comments […]

  14. Cybergate9.net, right Allen? It’s working fine for me. I’ll save a local copy of the webpage and if it’s still not working for you, I’ll email it to you.

  15. I log into SL for about two minutes every day or every other day to see if my land parcels have sold so that I can convert the Lindens to USD and be done with SL.

  16. Sigh,

    Here’s what I said again:

    You don’t judge a country by how many people are born and die each day, or how many come and go. That’s a MMORPG hangover.

    You just [judge] a country by its GNP, and you judge its economic health with things like “starter homes” or “home purchases” etc.

    It’s always so difficult for me to deal with the tekkie literalism I always find on this and other gamerz blogs, and the zeal to play gotcha. It almost makes me not want to post. In fact, if you’re going to be literalists, be literal about something intended as literal, “how many people are born and die each day” is a literal statement that doesn’t *have* to contain any emotional or speculative baggage about HOW they die — it just means that whether a country has half a million die or three million die per day, it doesn’t mean that the nature of the country per se is affected — there are good and bad country situations, and population isn’t always a marker. Furthermore, I said the all important qualifier “thinks like”. That’s not to suggest that home ownership and land ownership is the only marker of a RL (or virtual) country; but given that in virtual worlds, you can’t get AIDS, or cancer, or die from a car accident, the *nature* of your death isn’t at issue in quite the same way.

    Raph said:

    Judging a country solely by landowners is kinda one dimensional too. Knowing how many people are residents, workers, tourists, and so on is a valuable piece of data. For example, you yourself use it to arrive at your ratio of landowners to total visitors.

    Well, yeah. But we already *had* the statistics for literally who came and went. And my point was that *in addition to that* you need to get away from the MMORPG obsession with subscription numbers, logged in last 60 days, and concommitant log-ons and look at the stability as represented by landowners.

    That doesn’t mean that I’m an evil leering capitalist land baron who wishes to exploit teh ppl 1111 who only views my empire’s “health” as a legitimate index. Don’t be silly, if that’s what you’re implying.

    But how am I to distinguish from resident, workers, and tourists? Seriously, Raph, there is only one way, within the narrow, synthetic confines of SL: land ownership. A “resident” could map to the premium account with land ownership. A worker is…what? Somebody in a sex club? Or a top content provider who owns a sim for his content to be in a store? And what is a tourist? a day tripper or someone who never gets a job or land? There are venerable SL residents who never do either.

    The MMORPGized viewing SL is the one-dimension here — I merely suggested one corrective — and a very important one. Because if you unpack it, it contains a lot:

    o a landowner is willing to pay tier for 30 days in advance — stake.
    o a landowner is willing to sign up for premium account and give credit card and personal information
    o a landowner will likely terraform his land, landscape it, put up a house
    o needing a house, means buying prefabs, furniture, pools, etc. and stimulating the rest of the economy and feeding the content creators
    o a landowner doesn’t fly around and grief, or knock crap together in sandboxes or aimlessly wander around clubs. He stays home and likely has friends over; has a partner or even spouse with an intimate relationship; has projects he’s working on to either improve his home or business; has meetings at his home, etc.

    Land-ownership is therefore a good mark of stabilization and prosperity for a society. It’s not its only indicator of health but it’s a good one — if you have a country where the state is supposed to house everybody either in shanties or even in refugee camps, it’s not healthy.

    Now, again, to address the literalist problem, mapping the coming and going from a MMORPG to a RL-country is an incomplete mapping exercise. Why? Because they aren’t pixelated avatars but real people. So HOW they come and go is important! Like whether they die in infancy; in childbirth; from accidents; from preventable diseases; from war; even from genocide perpetrated by the government.

    But how were you planning on mapping that stuff to a MMORPG or even a synthetic world? (I suppose you could try…)

    2.5 hours ago, I got up and stepped out of an alumni lecture on fairness and factions in global health. Don’t tell me you can’t judge a country by how many live or die. Those rates mean a lot, not by themselves, but no MMORPG judges them by themselves, either.

    Uh, Michael Chui, spare me your third-worldism and moralizing, please — don’t tell me my business when it comes to judging either a real or a virtual society — I suspect I’ve got a good deal more experience in this regard than you have in either. And I assure you I speak as one who doesn’t just sit in lectures halls to hear about these things and wring my hands about them or press on an awareness button in a game or the Internet, but who has had a more direct, RL relationship to them.

    Honestly, the lengths to which people go to make facile arguments, trying to shame people into silence by making them out to be evil exploitative Westerners who are out to kill the developing world with their uncaring! Please. Juding countries — or games — merely by their births and deaths is not enough to understand what’s happening inside them.

    Countries are complex things to judge. So are MMORPGs.

    But there’s no QUESTION that virtual worlds being judged by these very narrow figures traditionally tied to MMORPGs are simply insufficient. In fact, maybe they’re even a poor metric to be judging MMORPGS by! Could people crack open their minds just a bit to entertain THAT thought? I remember they years of dire predictions that The Sims Online was dying and closing, dying and closing. But it persisted with a hardy core and it was kept open. This sort of thing should be studied and understood, not dismissed as statistically irrelevant.

    If you could just set aside the MMORPG lenses and the literalism for a second, you could hear my point:

    In SL, landowners doubled within a year, but remain a small percentage of the total, and a lesser percentage of the total than a year ago.

    Conclusion: for a business that has made its profit model based on selling islands or mainland sims off the auction, this is a sign of trouble that’s important to watch, underneath all the huge figures of signs-ons, that will likely reach a million in a week or so.

    Now, as to the spike:

    There was a major security breach on Sept. 6, announced on Sept. 8 — the dbase was hacked, and Linden Lab reset all the passwords. By the time people figured this out and found that their efforts to manually reset their passwords again were failing in some instances — AND had the additional factor of Sept. 11 being a school holiday for some due to memorial services for 9/11, that meant that you saw a huge spike in alts being made as people struggled to get back in at least on an alt. The Lindens made a heroic show of manning phone banks from then on to try to reset the passwords. There was more press of the major, mainstream, variety than there had ever been — and that actually caused a lot of people to come and try it, too.

  17. Looks like the link works this AM.

    From a literal data perspective and from the SL Web Page:

    These numbers don’t include profits generated from land-related
    transactions
    – These results understate profits for those business owners who spend part of their proceeds in-world for non-business-related purposes.
    – These numbers don’t include the significant proceeds that some businessowners receive from 3rd parties in US$ (paypal, check, etc..)

    Prokofy-

    Given these statements we can conclude (correct me if I’m wrong please):

    1. SL Tracks Total Residence
    2. SL Tracks Active Residence
    3. SL Tracks Residence Online
    4. SL Tracks USD Spent on a rolling basis by 24 hour increments
    5. SL then by defualt tracks summary numbers of all of these on a daily, monthly, quarterly, and annual basis.

    Further by the statements above:

    6. SL Tracks profits generated by the sale of land
    7. They control for in-world transactions, non business related, in other words they know a % of the money spent by business owners on non-business transactions or likely audit or spot check for it somehow.
    8. The numbers dont include the amount business owners recieve from outside sources (we can call these non-linden lab regulated markets)

    So if they are not releasing land sales figures (6), unless we know the frequency and average price of land sales we cant assume land ownership in and of itself is a good metric, it might be a significant source of revenue as you say, but thats information seemingly being closely held by LL.

    In addition we cant really get at (8) outside non-LL markets (paypal etc) because we dont have those numbers.

    However, it would seem that (7) we can extrapolate from the data available.
    I refer to what I wrte above:

    USD:
    Same as above, however remember the amount spent as a % of total population, active poulation and online population is a rolling 24 hour total.

    Meaning they’re making a mean of 50.12 per online resident, 1.23 per active resident and .54 cents per person in the total resident population every 24 hours.

    That seems to me to be a pretty vibrant online economy. Even controlling for not having the land sales figures and outside revenue numbers thats still very impressive. It proves SL has quite a bit of economic activity per redident logged in (likely premium account holders, and discounting tourists) but not so much for those outliers or those infrequently logged in.

    I’d actually like to see a distrobution of linden currency by account holder and account type, then you could actually look at things like concentration of wealth x amount spent while logged in or x frequency of login. Which would give you a good idea of behavior patterns by account wealth.

    PS: Thanks for the clarifacation on those days r/t the DB hacks, its likely good I wanst able to include those numbers after all.

    MikeRozak-

    Good analysis 🙂

  18. PS: Excuse my spelling/grammer…no caffine yet 🙁

  19. You know, I’ve made so much effort in the past decade to understand what people mean when they say things, and I’ve realized it’s all for naught. People don’t mean what they say. They have this alternative jocular meaning behind their words that I’m supposed to get, but I’m simply not cool enough to be in their elite squad of commentators. It’s probably just easier to assume you’re wrong, whatever it is you meant to say, but since I have no idea what you said, there’s nothing to argue.

  20. Back when I was in cellular/wireless retail, we had a few big numbers we had to hit. And one year, one of the big numbers we had an eye on was $600 million; the gross revenue goal for one of the states I was doing marketing for. And we had a certain number of new subscribers we had to bring on and a certain number of current subs we had to retain to hit that number, and the current ones had to keep spending at a current level, and some of them had to increase spending, and churn could only go up so much, etc. etc. Lots of graphs and charts. Fun stuff. If by fun you mean complicated, stressful and frightening.

    In one meeting where we were trying to make it all make sense, the sales VP for the state finally threw up his hands and said, “You know what I want? Instead of us trying to figure out how to get a million people to spend $600 a year with us or so… let’s just find one dude who will spend $600 million a year with us and really treat that bastard right!”

    We looked, but couldn’t find that guy 😉

    My point being that from a financial perspective, unless you can really connect the value of the various pieces of the SL puzzle… it’s a very, very complex game. For example, the landowners may be providing, overall, the higest percentage of revenue to Linden… but they may also be reaping back the most financial rewards themselves, to the point where they may not be out of pocket at all. So what? you ask. Isn’t that great? Isn’t that the point of SL? To allow exactly that wonderful content-creation space.

    Well… it is… for now. But if the majority of the income is coming from folks who are paying to be providers rather than players, then they are essentially reselling the product, to a certain degree. We hear again and again, for example, that Anshe is the biggest earner in SL, and that comes from land sales. OK. But if that land is sold to folks who then make money by renting to others who sell stuff or who sell stuff in order to pay for the land… you have what amounts to a pyramid situation in reverse; the smallest number of customers providing the largest volume of revenue to the publisher.

    I’m not saying this is the case; I’m just speculating that it could be. And when you’ll run into a real problem is when you get another alternative to SL that offers a similar or better end-user experience (the free stuff) but without the need for the mitigations of the folks who are now at the top of the content and revenue provision pyramid.

    I hope I’m wrong. I get a kick out of SL. But, as Prokofy points out, home ownership is a very important measure of economic importance. Why? Not because owning a home is per se better or smarter than renting, but because it signals and supports other economic activitis. If only a relatively few members/players of SL are supporting Linden with direct revenue, and they are relying on “downstream” players for their revenue and their satisfaction in the experience, then that is a double-choke on the economic model. In a standard subscription model game, you we say, “X players have paid the publisher A amount this month.” And “A” don’t change.

    In SL’s model, we may be looking at free or very low revenue (to Linden) players (p1) pay some revenue (r1) to landowners and retailers and designer level players (P2), who, in turn, pay their revenue (R2) to Linden.

    So… you could have a situation where, let’s say, ten P2-players are providing competing, inefficient products/services in a particular SL category or industry. They collect r1-revenue from some number of p1-players, all own land and pay Linden a good chunk of R2-revenue. In comes one new, very good, very motivated, customer-focused P2-player who wants to own that category. His product kicks righteous bootay. Soon, he has wiped out all those ten P2-players with his great new thing; it is an improvement, all agree, to the old varieties, and he charges about 1/2 the price of the average old product. So he’s making 5-times the r1-revenue for himself as the average old competitor, his customers are getting a great product at half the price… but he only pays about 1/10th the R2-revenue to Linden.

    There are a bunch of other scenarios where this “two tiered” system of economic interest gets really funky. And it might actually be a four or five tiered system, depending on how you consider land-ownership vs. having a paid account without owning/managing land, vs. working vs. providing designs, etc. etc.

  21. Residents Logged-In During Last 7 Days 112,918
    Residents Logged-In During Last 14 Days 150,554
    Residents Logged-In During Last 30 Days 235,311
    Residents Logged-In During Last 60 Days 351,206

    One important factor that hasn’t been mentioned is the fact that the total number of registered accounts has increased drastically over the time period of interest.

    For example, over the past 30 days, the number of accounts has gone up by roughly 215,000 (~620k up to ~835k). If one assumes every new account logs in at least once, then the 235,311 unique logins over the past 30 days is underwhelming from the point of view of SL’s long-term retention.

    Similar trends can be seen when one examines the 7 and 14 day numbers. In these cases, one finds that 64,000 and 60,000 accounts that were registered prior to the time window under consideration have logged in during the 7 and 14 day periods respectively. This indicates that of the 47,000 accounts that were made two weeks ago, roughly 4,000 also logged in last week. Again less than stellar from a retention POV.

    I don’t have my spreadsheet open, so these numbers are kind of rough, but I’m sure you see the point I’m making — when the total number of accounts is changing so quickly, it is necessary to normalize your data to account for that.

  22. You just a country by its GNP, and you judge its economic health with things like “starter homes” or “home purchases” etc.

    Here are numbers you need to look at to understand the world’s stability, and level of civilization:

    o number of landowners in June 2005: 6,000
    o number of landowners in October 2006: 19,000

    Hmm, Well, I’m not just worried about the economic health of the avatars of SL. Economists have allready calculated the GNP for Everquest and I’d be surprised if SL could do well verus Warcraft in terms of hourly wage if we count a gold piece as .25 US. I’m more concerned about how popular it is, how many people think it has value. I’m also looking for numbers that suggest Linden Labs may have a viable buisness model. But looking at your numbers, are you saying the game has about 25,000 real players (taking the 19,000 owners and adding a goodly sum of renters?)

  23. Ooh, and the money sink for Upload charges is three times that of land costs to Linden. Interesting.

  24. But looking at your numbers, are you saying the game has about 25,000 real players (taking the 19,000 owners and adding a goodly sum of renters?)
    # Rik said on October 4th, 2006 at 2:56 pm:

    Ooh, and the money sink for Upload charges is three times that of land costs to Linden. Interesting.

    1. No, I don’t take some narrow, superior, latifundista view that only landowners “count”. I merely point out that they bring stability to the world as they signal other types of major content buying of houses, skins, etc. but also payment of wages, purchase of other people’s land, etc.

    If 10,000 are logged on at any given time and Lindens tell us the average log-on is 4 hours, and that means 60,000 people a day logged on (another way to get at true numbers), and of these, 20,000 or 25,000 are landowners, or landowners plus renters, then that means the rest are flying around trying to figure out what to do. Going to clubs, events — and in some cases, griefing.

    Some tiny portion — 3,000? or 1,000? of those famous people who “make their living in Second Life” may be bent over their looms, sewing their asses off, trying to make content, or hunched over their lathes, tooling and dying objects like vehicles for the masses. They may do this in sandboxes or not even be logged in, and working PSP or 3-d modellers outside of SL. They may only own 512s. They upload like crazy — and frankly, high uploads aren’t even a sign of creativity or content generation of value, they are just a case of lots of new people coming in and doing at least one predictable newbie activity which is to upload their their own picture into the game to go on their own profile, or their own sample t-shirt slogan to upload on their own newbie t-shirt.

  25. If 10,000 are logged on at any given time and Lindens tell us the average log-on is 4 hours, and that means 60,000 people a day logged on (another way to get at true numbers), and of these, 20,000 or 25,000 are landowners, or landowners plus renters, then that means the rest are flying around trying to figure out what to do. Going to clubs, events — and in some cases, griefing.

    This is somewhat of an oversimplification, ignoring time variation of concurrency and variability of login durations, leading to an inflated logins-per-day number.

    If you plot up the 7-14-30-60 numbers, you can fit a nice power-law relation to the data ([total uniques in interval t] ~ t^0.53). Extrapolating that back to t=1 day, we find a total daily login number of roughly 38,000 uniques per day. This is about a third fewer than your initial estimate.

  26. […] Raph’s Website » Second Life posts weekly uniques Says: October 3rd, 2006 at 5:35 pm […]

  27. […] SLurlBuilder (secondlife url sl forau) Bookmark  [Discover] Raph%u2019s Website � Second Life posts weekly uniques (mmo secondlife stats) […]

Sorry, the comment form is closed at this time.