Welcome to Jester's Trek.
I'm your host, Jester. I've been an EVE Online player for about six years. One of my four mains is Ripard Teg, pictured at left. Sadly, I've succumbed to "bittervet" disease, but I'm wandering the New Eden landscape (and from time to time, the MMO landscape) in search of a cure.
You can follow along, if you want...

Tuesday, March 6, 2012

Snapcount

Notice: The following post, and all of its supporting data, were written and generated without assistance of any kind from any current or former CCP employees or management.

Last summer, I made a bit of a name for myself with a pair of posts that looked at data focusing on how many people are logged into EVE Online at any given time.  I called the posts "Not all curves are pretty" and "Some curves aren't".  And aside from my various player guides, the second post is the most-read thing I've ever written on this blog.  The two posts were statistical analyses of EVE's logged-in player counts over a period of about five years.  I looked at EVE log-in data primarily because that's the data that's most interesting to me.  While other MMOs concentrate on subscription numbers, EVE's "content" is really primarily the players.  The more people playing EVE, the better the game is and the richer the experience is.

So while the number of EVE subscribers is important, to my way of thinking, the number of people logged in is more important.

It's my understanding that "Curves" created quite a stir within CCP.  Certainly, it probably represented the first example of an EVE player taking CCP business data, normalizing and collating it using modern techniques, and producing an analysis freely available to the player base.  The data for the first post was provided from publicly-available PCU data.  The data for the second post was provided to me by a pair of anonymous sources; since their data matched both each other and the previously publicly available data, I ran with it.

The center-piece of the second "Curves" post was this chart:



It showed a massive decrease in logged-in players following the release of Incursion and a completely stagnant log-in curve after the release of Incarna.  As a result, it provided both players and CSM6 with a lot of political capital to show CCP conclusive proof that from a business stand-point, Incarna was not succeeding in its stated goals.

CCP was apparently surprised and unhappy to be dictated to in this manner.  At the end of "Curves", I promised that as long as I was provided relevant data, I would keep the players abreast of any changes in these numbers.  But I also made it clear that I was dependent on the data.  I guess I shouldn't have been surprised when the data flow dried up.  The public data source I'd used for the first post itself went off-line, and I didn't hear from my two sources again.  Between this and Fearless, CCP became rather manic about plugging "leaks" of information.  Whether this mania contributed to the death of the QENs around the same time, I can't say.

And for months, the data that I needed to generate an update for "Curves" was unavailable to me.  It left me with a bit of a problem: if I wanted to continue tracking this curve, how could I do it?

I'm happy to report that I've found a way.  This time, I'm keeping the "how" to myself to hopefully prevent my data being shut off.  But you can rest assured that the data source is publicly available, and I now have the capability to independently pull that data and generate analyses of EVE logged-in player counts whenever I care to.  The core of my ability to do this is a pair of self-written computer programs that I collectively call "Snapcount", after an American football term.  A week ago, every element of Snapcount ran successfully for the first time, producing its first report.  I linked the new data to my previous data, and here we are.

Now that I have a good source for this data, I'm keen to do some other analyses as well.  For instance, I have a hunch that the so-called "summer slump" is a myth and I'll be interested to see if the data bears this hunch out.  Watch this space.

In the meantime, this post represents the long-promised update to "Curves".  Let's start with the full-year 2011 chart, the update to the chart above:



As you can see, after the end of the previous chart -- the data for which ended on 22 August -- things did not improve in terms of numbers of players logged into the game.  21 August was a Sunday, traditionally the busiest day in the EVE week, and represented a good number of players logged into the game.  28 August saw this number drop by about 5%.  It continued dropping all through the remainder of the summer into autumn.  By mid-October, the decline was almost 10% total.  This almost certainly represents player unsubs.  The 30-day delay after the Jita riots of mid-July reflects that a large portion of these unsubs were likely players using GTCs or PLEXes to fund their play.

By October, the green 90-day rolling average line is essentially in free-fall, finally leveling off in a new normal of about 27,500 players logged in average, per day.

In the first post last July, I included a chart that displayed the relative success of each EVE expansion based on the number of players +30 days after the expansion was released, +60 days after the expansion was released, and a week prior to the release of the next expansion, a date I called "+Next".  I can now fill this table in and include the Incarna numbers.  Those numbers are remarkably grim:

ExpansionFeature+30D+60D+NextResult
RMR/Bloodlines  T2s, Carriers4.6%+7.0%+11.9%Success
Revelations IInvention, rigs  +4.3%-0.1%+17.6%Major success
Revelations IISov changes+0.8%+6.4%-2.6%Failure
TrinityNew graphics+2.4%+29.0%+19.5%Major success
Empyrean AgeFaction war+0.5%+7.1%-8.5%Major failure
Quantum RiseNano nerf+18.5%  +30.5%  +24.7%  Major success
ApocryphaWormholes+11.4%+10.7%+5.7%Success
DominionSov changes+13.6%+13.5%+6.4%Success
TyrannisPI+6.1%+2.6%-0.6%Failure
IncursionIncursions+8.9%+16.8%-5.0%Failure
IncarnaCQ+0.3%-0.2%-10.7%Major failure
CrucibleIterations+7.2%+13.7%??????


The record is clear: in terms of attracting new players and retaining old ones to play the game, Incarna ranks as the least successful expansion in modern EVE history.

So much for bad news.  Let's turn to the good, and it starts with the new line at the end of that table.  As you can see, in its first 60 days of release, Crucible has gone a long way toward repairing the damage!  So far, it ranks with Dominion and Apocrypha as a solid success.  Here's the updated chart showing average logged-in players for the last six months:


This time, concentrate on the maroon 30-day rolling average line.  After free-falling to its new normal of around 27,500 players logged in per day, Crucible was released on 29 November 2011.  And as you can see, the improvement was almost immediate!  After a dip caused by expansion-related downtime, things are heading back upward right up to the present day.

26 February was a Sunday, the last Sunday shown on this chart.  And if Snapcount is to be believed, 26 February was the most successful Sunday EVE Online has had since early June 2011, eight months ago.  Out of the woods yet?  Nope, not hardly.  The Incursion expansion set new records for numbers of logged-in players.  With the bleeding stopped, the current counts have recovered from the 10.7% dip inflicted by Incarna.  Things are back to where they were just prior to the release of that expansion.  Prior to Incarna, looked at in the long term, EVE's logged-in player 90-day rolling average was stagnated at 30,000.  Right now, we're just below that.

But there's certainly cause for hope!

And thanks to Snapcount, I'll be able to keep you updated on it into 2012 and hopefully beyond.

39 comments:

  1. Good post. I think it would be interesting for someone to see how closely the published sub numbers over on mmogcharts follow the online user counts.

    ReplyDelete
  2. Start the 90day, 30 day lines at the correct point rather than showing them tailing down. This is just wrong and can lead people to misinterpret the graph. (Sorry statistics man it just annoys me. Year month of launch of the expansions on table ? And the mini-iterations ? The drop before Incarna is almost exactly the same as that after. Why did so many leave after Incursions? Random thoughts

    ReplyDelete
    Replies
    1. Incursion brought lots of players back to try the content but as the expansion didn't really deliver much else than a new PvE feature and the new character creator, it did very little to address EVE's biggest issues; PVP.

      Incursion didn't address the ship balance issues that were stagnating null sec or the pointlessness of FW. There was the little things project which had good stuff in it, but not much that shook up the PVP aspects. People tried incursions for a bit, the novelty wore off, and they logged off again.

      Delete
  3. I could read these all day... GJ!

    ReplyDelete
  4. I also don't think your +Next is the best way to judge success. A week before the next expansion it is already known what is going into it, and for the good expansions, all the pre-expansion hype and marketing will make the number artificially high. I would expect the numbers 90 or 120 days after the release to be more representative (even your 60 day numbers look closer to the generally accepted views for most expansions)

    ReplyDelete
  5. This seems like conflicting information at first.

    Your analysis suggest we are almost back to Incursion expansion levels, but CCP announced in the last interviews that they had a net growth in the year.

    I can only conclude that the PCU/subs ratio is decreasing. I wonder if it is because people are getting more secondary accounts, but not playing just as much with those.

    ReplyDelete
    Replies
    1. I could subscribe for up to a year in one payment, and be counted as a subscriber for every month of the year. But I might lose interest and stop logging in during that period.

      CCP's growth figure would count my year long sub as a +1 for the year, but Ripard's figure would show if I stopped logging in after Incarna was released as the PCU drops by 1.

      Also, although CCP have sustained growth, we don't know by what margin. Also, the incursion feature wasn't released until January as the expansion was broken up and staggered. It may be that some of the new subscribers generated by Incursion crept into the beginning of 2011 rather than late 2010 when the expansion was officially released.

      Delete
  6. The Australian time zone has more players now and when the Japan client hits it will have more.

    ReplyDelete
  7. Interesting data, and good writeup, as always. I think I know where you're getting the numbers from.

    Incarna shouldn't be considered an expansion, at least in my view, for two reasons. (1) Its core "content" -- the new avatar system, was released as part of Incursion's 3rd Step. (2) It was a contentless expansion, because neither new avatars nor captain's quarters provide any meaningful content besides a few hours of eye-candy. Incrana was essentially a technology demo. It should be considered as such, whatever CCP may claim.

    Also, I wouldn't use the MA from the week before the launch of the next expansion to judge the success or failure of an expansion. First, because the MA probably incorporates player activity following the expansion release, which is going to be high, unless you're using a 7-day MA or something. Second, because as Two step pointed out, pre-expansion hype might inflate player activity; we see such behavior for previous expansions. Third, events occur following an expansion's release that can have a big impact on player activity. The Sanctum Nerf probably had a big impact on player activity, but because it came following Incursions your chart attributes the falloff in players to Incursions, even though the two were not related. Similarly, Incarna in and of itself probably didn't impact player activity nearly as much as did the Summer of Rage.

    Probably the best point for judging an expansion's success at increasing long-term player activity is the 90- or 120-day post-launch weekend.

    ReplyDelete
  8. One month before the next expansion would be better, to avoid counting people logging in and buying up products that are going to become scarce in the next expansion as "players" :P

    Also, the sanctum nerf happened during FanFest. Note that the numbers dived for a short period while about a thousand players were in Iceland, the popped right back up.

    ReplyDelete
  9. "CCP became rather manic about plugging "leaks" of information."

    I have the impression that CCP won't disclose subscription count during the next fanfest. Probably neither the employee count.

    ReplyDelete
  10. Love the analysis, Jester. I would also be curious to know what the PCU/subscription rate is, but oh well.

    Should be interesting to see where Inferno takes us with FW and wardec fixes. I'd say especially so regarding FW which didn't seem to do so well even when it first came out (Empyrean Age was your only Major Failure besides Incarna).

    ReplyDelete
  11. Love the post and your efforts to gather the information. Do you consider the new client resource requirements might have had any impact on these numbers? When Incarna was released I went from being able to run 3 clients concurrently on my PC to only 1. While that is now back up to 2, I would say my accounts still spend half as much time logged in as to what they used to, just due to the memory needs and the impact they have on other programs running on my PC.

    ReplyDelete
  12. Jester, I often disagree with you and have made several comments to indicate that. Posts like this are probably why I dislike you most. You made perfect sense and used hard numbers. I makes if very difficult for someone like me who regularly disagrees with you to dislike you personally. Further it really makes me feel like an ass when I don't agree with you and I begin to question my own perspective on those occasions.

    It would be helpful to me if you could come up with a nonsensical asshat post filled with untruths to make me feel better about myself.

    ReplyDelete
    Replies
    1. Ugh, I'm sorry. I'll try to write something that infuriates you tomorrow. How would that be?

      Delete
  13. If I was an investor thinking about a buy-in (or further loans as a bank officer), I'd have wanted to see this kind of analysis written up by the CCP management and presented with a business plan rather than see it posted by a third party named Jester. What I find amazing is that this type of analysis could not have been used by CCP management (who should be way ahead of the customers/bloggers) to gauge what was happening to their business long before the actual release of Incarna and have suggested the need for a plan "B", especially considering the CSM feedback they were receiving.

    If I had to make a bet I would guess that the revenue stream for CCP from EVE is almost an exactly proportional fit to the curves (or not curves) you've posted here. And that type of analysis would show that before CCP does any IPO they would have to demonstrate the ability to recover above that 33,000 to 34,000 mark at a minimum. IPO's aren't normally initiated (at least not successfully) by companies that can't show sustained growth.

    I applaud your work - it's excellent --- but in a sense it really calls out even more the failures of the CCP business management over the last 18 months. This kind of metric is the kind of real-time indicator CCP should have been paying attention to.

    ReplyDelete
    Replies
    1. Well, if they plan to float in the US, this is the sort of analysis that investors will expect to see in their detailed SEC filing.

      Delete
  14. Doesn't EVE Offline have a graph like this running back to 2006? It's had one as long as I remember.

    ReplyDelete
  15. Why all the cloak-and-dagger stuff about the "sources" for your numbers? Haven't PCU's always been available through the usual API query?:
    http://api.eve-online.com/server/ServerStatus.xml.aspx
    Is snapcount a script which polls this info every hour?

    I assumed that's what eve-offline is doing (as well as numerous other sites), and their numbers look to be fairly complete over the last year.

    Otherwise, I guess you could just scrape the EVE login screen, where they've never been shy of reporting the PCU.

    Did the API interface go down for some period?

    You seem to be implying an effort to suppress data on CCP's part. But they've always been up-front about the numbers on the login screen, so that doesn't quite seem consistent.

    And let's be honest here: Although you justify why PCU numbers are interesting, the fact that is you look at PCU because those are the numbers which are available. Paid subscriber numbers would also be very important in understanding the success/failure of an expansion, but you don't analyze that because you just don't have that data. Your opening paragraphs make it sound as if you've made a choice, but actually you're just working with the numbers you have - And which we can thank CCP for being very open about providing via a convenient interface.

    ReplyDelete
    Replies
    1. Yeah, thanks for that comment. I was starting to wonder about that while reading....

      Delete
    2. Jester, do you have any additional information about why you believe CCP was suppressing PCU data as a result of your blogs last year?

      It's a strong implication. Your article is heavy with innuendo about how difficult it is to get these numbers. But there is ample evidence to the contrary, and I don't know of any supporting it.

      As you know, the data stream is the EVE-API (provided by CCP). It is a very convenient and reliable interface to the server which is at the heart of a number of amazing services such as Dotlan. It has never been the mystery you are trying to cloak it in. There are dev-blogs about it.

      If you got your historical data from a 3rd party, then they surely got it from the API. You know that because you checked your sources, right?

      And if 'snapcount' isn't using the API, it probably should be. Otherwise many who depend on the API would be interested to know why you think another source is more authoritative or reliable.

      These data are the basis of the blogs which made you eve-famous.

      So why the cheap-shots against CCP who have always provided them openly and dependably?

      Delete
  16. I won't argue that the logged in players aren't the most important data because they are. But you should add a curve with relation from (players logged in) to (accounts subscribed). There surely where a lot of unsubs but in addition to that there were (and are) Incursions running all over the universe.

    What does that change in Player behavior? If you have two or more combat chars you most likely fly sanctums or havens with both of them. Maybe you even had a third char salvaging after you.
    Did you ever tried to dual box incursions? I did it on Sisi and it is pretty damn hard to maintain for more than an hour.
    So the drop of players online will be partially be side effect of the good things incursions brought to us. People logged just in with there incursion char to make their money instead of logging in all of there char to do what ever they did before incursion to get money.

    But still a very good post. Keep the data flowing who ever you "public source" are ;-)

    ReplyDelete
  17. I've been logging into Eve, on average, of 5 days a week for almost two years. I can remember a time when the devs were bragging about some 68K or 69K players logged in at once. I have noticed the numbers dropping, since the average for my time zone (+1 Eve) use to be around 40K during the week, and 60K on the weekends. Now it's just getting back up to 25K during the week and 45K on the weekends after having fallen to the low 20s / high 30s.

    ReplyDelete
  18. Awesome!!
    10/10!


    Please pretty please show us the full graph from day 1 (I dont care if the graph grows into a .png of 1,000x10,000 pixels... :) )
    Highlighted with the major patches, buffs, nerfs, releases, etc.

    I would drool over that! ;-)) That would give me and others too the ability to see the big picture.


    ps.: you should write a blog about some interesting things in the data, like the lowest loged-in day(s), the highest, the most tranquil and so-

    Cheers and keep up the good work!

    ReplyDelete
  19. Sinclair FergusonMarch 7, 2012 at 6:53 AM

    A great piece, as always. I'm curious how or even if the release of the long awaited and hyped Star Wars: The Old Republic (or Skyrim, for that matter) had anything to do with declines in the latter part of Q4. I know that I was disenchanted with Eve late last year, and unsubbed after 3 years of continuous play to give SWTOR a chance. Just a thought.

    ReplyDelete
  20. Excellent work Jester.

    I've been a player for about 4 years exactly. For the longest time, I've been infatuated with EVE and played daily regardless of the expansion or what it brought as baggage. I viewed Incarna with the same "whatever" that any other expansion got. The game is beautiful. It's spaceships. Therefore, I play. In the last year, my kids (wanting to do stuff with the ol man) got accounts. (which I pay for...how the hell did that happen?) I'm now up to 3.

    My time in game has decreased quite a bit since family IS more important but I still pay for 3 accounts. CCP says thats growth. While paying for 3 accounts is certainly growth from THAT particular spin, if I dont enjoy the content and stop playing, CCP can still see that as "growth". All hail the might spin doctors.

    Keep 'em honest, Jester and good work. My kids are reading over your Guides as they want to start into the pvp aspect of this game.

    ReplyDelete
  21. Your graph output isn't very good, and rather grainy. I would be happy to help you correct this. Shout if you want the assistance.

    ReplyDelete
  22. Good stuff Jester, although there is one data point that you don't address. The banning of bots in large numbers would affect the PCU graph far more than the subscriber graph and you don't seem to acknowledge any of the mass banning dates. I know CCP has become rather tight lipped lately, but I'd be interested to see if you're not misinterpreting any of the PCU drops caused by a more aggressive security team over the past year.

    ReplyDelete
    Replies
    1. I was curious if someone would bring this up.

      Short answer: Unholy Rage showed up very clearly in the data. It was quite obvious that there were hundreds fewer clients logged into the game on a daily basis.

      The latest bot sweep... ummm... not so much.

      Delete
  23. I took your graph and zoomed in to a particular date in August. I had to zoom in quite close. In fact, I zoomed in so close that the line became a series of contiguous pixels.

    Thats when I saw it! One pixel dipped down from the pixel on either side of it. That was me! I was there!

    ReplyDelete
  24. Jester, this post was terrible.

    The entire first half is self-aggrandizing nonsense. The PCU numbers have always been among the easiest data to collect from the Eve server - how to do it is documented (by CCP themselves) on Evelopedia. This source has never "dried up", certainly not for the political reasons you allege in a paragraph that wanders deep into tinfoil-hat territory.

    The second half of your post was scooped by Tippia more than three weeks ago. The 3-weeks of extra data confirm the trends he observed, but add little extra insight.

    Tippia's post has the added bonus of being honest about its source (the EVE-API) and free of the "Lone Voice of Truth" delusion that we read here.

    ""I'm happy to report that I've found a way. This time, I'm keeping the "how" to myself to hopefully prevent my data being shut off.""

    Good grief.

    Stick to the numbers.

    ReplyDelete
    Replies
    1. I would like you to publish something your self as so many seem to know the truth or know how to. The data on log screen could be easly manipulated. You all know that CCP is different than any other company and won't lie about the numbers because....exactly they would lie like any other company just so they would get invesor from outside to finance their projects, since they can't finance them on their own...any more.
      Another point...I know people are droping subs because I used to have friends playing with multiple accounts (I just droped 3 subs to keep only one, played all 3 for 3 years) and now most of them are gone and we were never going to leave...but we did. So did many other people. Stop bashing guy for trying to show that maybe it isn't as great as the "big guys" say it is. You don't have to agree...but I will take Jester's side before I will take some of you anonymous....especially the ones that sound like they come from CCP...yeah you know who you are...

      Delete
    2. Still doesn't make what anonymous number one said false. PCU numbers ARE the easiest data to collect from the EVE server, and if ever there was a time for CCP to falsify that data, it would have been after Incarna. CCP hasn't had a campaign to hide THAT data.

      All of Jester's tinfoil-hattery about 'getting the data' is basically just Jester trying to look important again. And he was, indeed beaten to it by Tippia by several weeks, and really nothing in this blog adds anything to that first post. This is just Jester, again, trying to look bigger than he really is, desperately trying to get attention again. If he'd just stick to the numbers this would be just a redundant blog. With all the tinfoil-hattery it sounds a bit sad.

      Moreover, this ignores completely the upward-downward trend we always see when an expansion comes out. A 30 or 60 day rolling average smooths that out, but it fails to do so at the right-hand side of a graph. It is, afterall, just looking backwards. Crusible is looking better than some, but we only know how much just before the next expansion hits on how far its effect has gone down again. At any rate, after the previous fail expansions, it would be a surprise, to say the least, if this expansion didn't have some positive effect. Especially given CCP's all out effort to clean up appearances.

      Delete
    3. I have to admit, I'm finding this conversation faintly amusing.

      I'll bet after someone bragged that they'd killed the first Titan without the use of any capitals or super-capitals, there was someone there that said "Oh, ANYBODY could have done that. All you need are 600 sub-caps, a good FC, tacklers, intel on the Titan, people to keep an eye out for cynos, and a few other things. Anybody could do it."

      Fine. Great. Any fool could do it. Sure. So why didn't YOU do it?

      Convos like this are the sort of thing that Mittens responds to with cynical mocking and lots of tildes, and from time to time, I really understand why. ;-)

      Delete
    4. Jester, I don't particularly care that "anyone could do it" - though they could. People have been looking at PCU numbers since forever and with more depth (check through the forums). If you want to act as if you invented that idea, that's up to you, no problem.

      But you are implying dirty tricks by CCP in their blog. But there is no indication that they actually played those dirty tricks.

      But CCP *are* providing your data, and have never suppressed it as you allege.

      I object to the fact that you have fabricated a narrative to build up your rep, while falsely trashing CCP's.

      This is a popular blog. People may be referring back to this article for months to come. And they are going to read a story about a paranoid company that tried to shut you down - and it's all false.

      You know that, and you should correct it.

      Delete
    5. Lol...Jester I think you have officialy +1 hater...as far as CCP and subs go...We shall see in the near future how far EVE will get. CCP dont F this up. Peace.

      Delete
  25. "Fine. Great. Any fool could do it. Sure. So why didn't YOU do it?"

    I don't have to - Tippia already did it for me. (so did eve-offline, for that matter, if you don't mind holding a ruler up to the screen.)

    And he stuck to the facts and was honest about his source, so his post didn't garner any complaints.

    ReplyDelete
  26. So. How is (UI)nferno going?
    Looks to me like it's lost CCP 6-8000 daily average users.

    Does that make it a 'Stupendious failure'?

    ReplyDelete

Note: Only a member of this blog may post a comment.