Itâ€™s been almost a year since our last effort to measure the webcomic audience. While in a perfect world we would have spent that time developing proprietary measurement tools capable of providing a highly accurate list of webcomic audience numbers this, in so many ways, is not a perfect world. Plus, we spent the development money on Mexican vitamins. But that alas is another story.
Way back in November of 2003, we published the last version of the Most Read List where we relied on data from Alexa.com and Ranking.com. Both sites rely on surveys of traffic that can lead to unpredictable skewing of the results. Alexa.com in particular relies on data obtained from the Alexa toolbar that users must download and install on their computers. Although this is not as accurate as the systems used to measure CD or book sales it is actually probably no better or worse than the diary system for television used by Nielsen or for radio by Arbitron. So while we will continue to investigate methods to arrive at a more accurate Most Read List, we thought there was a value to providing an update using the current approach.
For this list we looked at Alexa rankings and Ranking.com rankings (based on a combination of a sort by page views and a sort by visitors). We specifically looked at data from the month of September (through Thursday, September 16th) for the initial effort to rank webcomic audiences. We also reviewed Alexa data for the last three months, however, to make direct comparisons between webcomics on the list after the initial sort.
Penny Arcade once again leads the chart, which is either evidence of this webcomicâ€™s continuing popularity or an unholy alliance between Alexa and Gabe and Tycho. And not surprisingly, after a highly visible year for creator Scott Kurtz, Player VS. Player has moved up to the number 2 spot. A number of comics make their rookie appearances on the list including: Ctrl-Alt-Del, Angel Moxie, VG Cats, Least I Could Do, PartiallyClips, Kevin and Kell and General Protection Fault. A number of other webcomics just missed the cut including: Goats, RPG World, Wapsi Square, Itâ€™s Walky, El Goonish Shive, White Ninja and Get Your War On.
As always, this project is in beta so please provide feedback on any aspect of our efforts. We have also continued to add webcomics to our watchlist. As before, this list only ranks those webcomics actually on our watchlist. To the extent that we are missing a webcomic on our watchlist that potentially would place on our most-read list, we rely on you, the readers, to bring it (them) to our attention.
|Rank||Name||URL||Last List||Time on Chart|
|2||Player VS. Player||pvponline.com||8||5|
|18||Least I Could Do||leasticoulddo.com||–||1|
|20||Bob and George||bobandgeorge.com||7||4*|
|21||Kevin and Kell||kevinandkell.com||–||1|
|23||Scary Go Round||scarygoround.com||24||2|
* Non-consectutive appearances.
You can read all of the webcomics on the Watch List here.
Well, this list is TOTALLY inaccurate. Looking at Keenspot logs, the strips from keenspot are in the wrong order, and strips that SHOULD be there as they are more read than the ones on the list.
What’s the point of this when you KNOW it has SERIOUS inaccuracies?
ummm… to have an -approach- to the most read comics around?
Let’s give them a little of credit, shall we? After all, let’s not forget this is a beta thing, as they mention. They use the tools they have at their disposal, and that’s not too much, after all. Besides, let’s not forget that this isn’t an extensive list of ALL the comics around, but the comics that people like you or me suggested to the list.
Anyway, I’m interested on this project, and would love to see how it evolves. I’d like to add my two cents if I have the opportunity, too. 🙂
Scrubbo – once again you’re TOTALLY exaggerating to make a point. The methodology is laid out pure and simple in the article. Keenspot has not offered to share their server logs with us (granted I haven’t point blank asked the keen CEOs for them but I would certainly be willing to talk about it with them if they were willing). Frankly, although I was surprised by some of the entries at first, I think the list continues to be useful because it provides an idea of what the most read comics are. People bitch about tv and radio ratings all the time. People used to bitch extensively about the Billboard charts before Soundscan took over. We don’t have a Soundscan for webcomics yet.
I, we all, are happy to improve the methodology of the list. Hell I’d be happy to just fix the list if it’s as inaccurate as you claimed. So let’s hear it: what are the server stats (page views and uniques or sessions will do) for all Keenspot comics for the period in September measured?
I just posted this over at Keenspot as well–
Scrubbo and Jim – I’ve already answered ScrubbyDub on the Comixpedia Site.
The URL for the article is
Jeff’s comments are very interesting – Jeff could I repost them over at Comixpedia on the article’s thread.
Really why do this? Because I get this question ALL THE TIME. Why the list as it is? Jeff nailed most of it on the head. If Keen or anyone else was willing to share some numbers, even summary numbers we might have another way to do this. Although than I run into the trust issue – how do I verify it’s true? Although third party counter/clickers are more inaccurate at least I know it’s a third party and they should not have the same possible reasons to cheat.
One thing I toyed with earlier in the year that I could return to is this idea: Everyone who wants to play in the survey puts a Comixpedia counter on their site for a week to give me data on their sites. It’d be like a “sweeps week” for webcomics. Granted that too raises problems – people would probably do all kinds of silly stuff that week to raise their numbers. Still I wager that might provide a fairly decent way to order webcomics.
Jeff Darlington agreed to let me cross-post his comments on the List on a Keenspot thread:
First Post from Jeff:It might help to link to the actual article, though, so folks don’t have to go digging for it. (Especially later, when there’s no longer a link from the main Comixpedia page.)
One minor correction: GPF has actually been on the list twice; it was on the very first version. Since you’re counting non-consecutive appearances for other comics, I thought I’d point this out.
Now, on to my biggest problem with the list: There has got to be a better way of compiling such a list than to rely on spyware (i.e., Alexa) to get the data. This has been my biggest quirk with the list in the past, and unfortunately it’s still there. I know for a fact that my ranking must be lower than it should be, simply because my readers are probably smart enough (or they should be smart enough) not to expose themselves to spyware like Alexa, and I’ve personally strongly discouraged it to those who have asked.
Of course, there in lies the main problem. How do you get such data in the first place? Certainly, the most accurate list that could be compiled would require the artists/webmasters to hand over the raw server logs (or at least aggregated results like those that Webalizer produces). Unfortunately, I don’t think very many people would really go for such a thing–even I’m leary of giving out more than summary numbers to close friends and colleagues, and I know a number of artists who are far more paranoid about that than me. In addition, some people simply don’t have access to their server logs (I didn’t before I moved to Keen), so they may not be able to provide them. Third-party tracker images can do in a pinch in those cases, but they’re a lot less accurate. “Top X” list sites like the old World Charts Global 100 or buzzComix are pointless because (a) they rely too much on readers to “vote” for strips, which only (b) drives traffic to the ranking site and does nothing to really promote the sites the readers voted for. All those do is prove how gullible your readers are.
I noticed in this related article from last year that you asked for ideas on how to better compile the list. Unfortunately the link to the Comixpedia forum thread doesn’t work any more. Did anything good come out of that?
Second Post from Jeff: Well, I was trying to be helpful and spark discussion. After all, all of these flaws have been brought up before; Xerxes and I went back and forth for a little while when the list first started, discussing the same problems. I’m sure that if enough of us put our heads together, a better, more accurate, more comprehensive, and hopefully automated system could be developed. But I don’t think it will really happen unless all cartoonists are onboard and are willing to share their stats, at least one a central, trusted collator.
On problem I immediately see when I look at the numbers Alexa provides is that they’re all rankings. There’s no hard data available, so it’s difficult to compare the Alexa data with the raw server logs to see how close they are. (At least I’m not seeing any raw numbers, but I admit I could be looking in the wrong place.) It’s also impossible to know exactly how the ranks are devised, since the raw numbers are not presented. If GPF is ranked 83,992nd out of all sites Alexa has surveyed by spying on their users, I’d like to see the hard numbers behind that calculation. Without those numbers (and in the case of “reach,” all the other numbers they have available), I can’t compare them with my own numbers.
The closest thing I can really test is pageviews per user. I can look at my raw server logs and take my real pageviews and “sites” values (sites being individual IP addresses, the closest measurement to actual readers) and come up with approximately the same calculations, then compare those to the Alexa numbers. Of course, there must be some amount of error assumed in the comparison, since I can’t see Alexa’s calculation formulas or their raw data, so I can only guess as to how they arrived at their numbers.
PAGEVIEWS PER USER:
Today: ALEXA: 4.0; GPF: 4.822763662
1-week average: ALEXA: 3.6; GPF: 5.813503427
3-month average: ALEXA: 2.4; GPF: 5.608617109
3-month change: ALEXA: down 17%; GPF: not sure I understand their calculation
In this comparison, Alexa skews way lower than what came out of my spreadsheet. When you compare that with, say, the other server logs on Keenspot, the disconnect is blindingly obvious. As already pointed out, there are several Spot comics that have higher traffic (by raw numbers) than the Spot comics listed, but aren’t on this list.
Another factor has to be accounted for is how actively the artist encourages the use of Alexa. For example, as I already stated, I actively discourage it’s use. I firmly believe Alexa is spyware, and I don’t want the information it collects to be spirited away from my machine without my knowledge or concent. However, Howard Tayler actively encourages the use of Alexa; he displays the pre-fab Alexa ranking graphic on his main page, for all to see. Howard’s Alexa ranking is 24,711 higher than mine (59,281), and thus comes in higher on this list. However, for September-to-date, Schlock averages about 65% of GPF’s readership. Not putting down Howard in any way, but that seems obviously off from Alexa’s numbers. (On an interesting note, Howard does average a higher pageviews per reader number than I do. His one-week average is 8.512077436, compared to my 5.813503427. Thus, he gets his readers to click on more pages than mine do, although our average pageviews are roughly the same.)
Well, at least it gets the dialog started. Hopefully if people become more aware of it and become concerned about it’s accuracy, better means of tracking will arise. As I mentioned to Al Schroeder in a thread on Buzzcomix, I’d be inerested in ultimately seeing analysis and surveys that give a sense of demographics and trends. (Not that I’d have any idea how to do that mind you.) Over time this could really be useful in charting the development of the medium.
P.S. It’s not likely to be a real issue any time soon, but I noticed that I’ve been added to the watch list but under the old address when I was hosting my strip on my portfolio page. It’s just Acid Keg at http://www.acidkeg.com/ these days. If you could change that some time in the next couple of years that it takes me to begin to scratch the bottom of the top 25 it would be appreciated. Thanks!
I’m not totally exaggerating. This list has had the same problems since it’s inception and NOTHING has been done to improve it.
If you actually DID do the “add a comixpedia tracker to your page” thing, then you really would have a fairly decent chance of providing an accurate list of participating comics, and while that system would be flawed because it wouldn’t have a lot of the ‘big name’ guys because they wouldn’t bother to add the tracker, it would at least be more accurate. Your other option is a browser plugin that generates pageview data like Alexa that comixpedia viewers can install to participate in the survey. If you made the code open source you could get around the spy-ware paranoia a little bit, perhaps, and have a chance of getting a good chunk of the webcomic reading audience to use it. Of course, you’ll still get skewed results, but the skew would be based on who YOUR readers are, and since the poll is FOR your readers, it’d be a little less arbitrary.
But relying on Alexis and Rankings.com is like CBS saying “Even though the documents are forged, we stand by the reporting” when you say the list has much validity, as is evidenced by Jeff’s post.
You’re prolly never going to get access to server logs from most people, so you’re never going to have a perfect system, but the only way to get one even remotely useful is to do it in-house.
I’m not going to keep repeating this forever, but of course you are exaggerating. This is nothing like “forging documents.” Nobody has been anything, but clear about how the list is compiled and where the data behind it comes from.
The list is not “totally inaccurate.” The webcomics on the list are among the most read webcomics, even if not under a “data from private logs from actual servers” system–actual top 25, in order, most read webcomics. These are very clearly among the most read webcomics and therefore help to answer the question what are the most widely read webcomics.
Again – show me the Keenspot logs for September and I’ll certainly take steps to use that information to improve the list.
Or just keep up the entertaining dialogue here. Either way I guess does no one any harm…
I’m not claiming it’s forged, I’m saying you using data you know is inaccurate is SIMILAR to them saying “we stand by our report even though these documents are fake”. You see? I’m not saying you forged things, I’m saying that using flawed data and standing by it is… pointless. Why make a list you know is inaccurate? Does it help your readers? It’s not really informative. Neilson ratings are important (if flawed) for advertisers, mainly. What purpose does your flawed list serve?
Then again, I pretty much would say the same thing no matter what method you used. I’m just saying you should get inaccuracies that are at least a little more germane, by having your readers answer surveys or having your own in house tracking or something.
As for the server logs, you’ll never get your hands (legitimately) on the logs of every keenspot comic because a bunch of the authors wouldn’t want it spread about. *shrugs*
Then again, I pretty much would say the same thing no matter what method you used.
Well at least you’ve cleared that up.
Well I can sort of help you out as far as Keenspace goes. Last month (August) I had 2113058 pageviews in total and my comic draws a little over 9% of all Keenspace traffic. You should be able to use those two numbers and the pageview rankings in the guide to figure out the exact number of pageviews any comic on Keenspace gets.
How to compair that information against other comics, I don’t know but at least you’ll know how things stand on Keenspace itself.
Comments are closed.