Ruffini proves that Kos’s traffic is inflated?

posted at 6:37 pm on October 4, 2007 by Allahpundit

A follow-up to yesterday’s post for the five or six people who care. After I and a few others questioned his theory, Ruffini conducted an experiment on Hewitt’s site to see if he was right. Result: he was able to generate two separate visits in the span of a little more than three minutes, seemingly by simple virtue of the fact that more than 100 other people visited the site during that span. SiteMeter claims that it counts 30 minutes of activity by the same IP on a site as a single visit so Ruffini would seem to be right. Except I’m not sure what he means by this:

I clicked back to Hugh’s blog at 4:36:43 PM, after I had been dropped. This resulted in the following new visit (#17,433,195) being counted, a full 26 minutes, 56 seconds short of Sitemeter’s claimed 30 minute window:

It sounds like he navigated away from Hewitt’s site during that three-minute period and then came back — which would, certainly, make the second visit a separate visit. (SiteMeter doesn’t specify that in explaining its 30-minute metric but I’ve always assumed that if you navigate away and then load the page again that it’s a new visit.) His theory is significant only if separate visits are being generated while users are navigating within a site, i.e. if someone clicks over to dKos, reads the frontpage for 30 seconds while 100 new users are flowing into the site, and then clicks a link to another page inside dKos. That shouldn’t generate two separate visits but if Ruffini’s right, then it will. I can’t tell from the way he worded his post whether he did that.

The only way to prove this on our end would be to compare our SiteMeter stats to the other stat programs we use. MM runs two others, I think; she’s never said anything to me about the numbers being grossly different but I’ll check. Whatever Kos’s actual numbers, it’s safe to say that he’s as big as Instapundit and MM combined. Which is pretty darned big.

Update: Ruffini updates to say he did not navigate away from Hewitt’s page. Like I say, I’ll ask MM about our stats.

Breaking on Hot Air

Blowback

Note from Hot Air management: This section is for comments from Hot Air's community of registered readers. Please don't assume that Hot Air management agrees with or otherwise endorses any particular comment just because we let it stand. A reminder: Anyone who fails to comply with our terms of use may lose their posting privilege.

Trackbacks/Pings

Trackback URL

Comments

( five or six people who care )

That many ? C’mon, your not fooling me. : )

Texyank on October 4, 2007 at 6:43 PM

Sooo, for the ill-informed, when a page counts “unique visitors” does that mean counting only page views from a given IP address once, no matter how many times that IP address navigates to and from a site?

See, I have this friend that clicks on and off of Hot Air about a million times a day, not with the intention of inflating visitor stats but due to some sort of addiction, and I’m wondering how that would affect both this page’s counter and the counter at Kos, if the same thing were done there.

NTWR on October 4, 2007 at 6:43 PM

No matter how big a zoo may be, it still is just a place that animals hang out and make noise.

And just because it is big, does not mean that it is not filled mostly with emptiness (empty heads, empty logic, blogs empty of rational thought, etc).

Voidseeker on October 4, 2007 at 6:46 PM

What about Port Address Translation? If a site gets multiple views from a single IP during a short period are those counted as individual views? What about persons navigating within a site using a single IP? (diferent people going to diferent places) Site meter needs to answer some questions.

Theworldisnotenough on October 4, 2007 at 6:57 PM

Interesting. Looking forward to additional data.

Makes me wonder if M’s new “live” capability could be used, somehow, for further experimentation.

T J Green on October 4, 2007 at 6:59 PM

This theory of Ruffini’s is sort of inside blogging, you might say.

Most important — in terms of DailyKos traffic’s value to its advertisers — is what might be called “the hover factor.” That is to say, diarists and frequent commenters who return to the site several times daily to see who has commented on their diary, or who has responded to their comment.

If Kos is counting such visits as “unique” — when in fact they are multiple daily visits by a relatively small hard core of devoted Kossacks — then that translates to a HUGE inflation of the DKos audience.

And I’m pretty sure that’s the situation.

Ali-Bubba on October 4, 2007 at 7:12 PM

It sounds like he navigated away from Hewitt’s site during that three-minute period and then came back — which would, certainly, make the second visit a separate visit.

Not necessarily. Not that it matters, I was able to reproduce his result (without navigating away). I don’t know that it’s conclusive proof, but it is certainly an unexpected result. I’m going to talk to SiteMeter and see if I can get some clarification.

Mark Jaquith on October 4, 2007 at 7:20 PM

Ali-Bubba on October 4, 2007 at 7:12 PM

Actualy looks worse than that…

Looks as though sitemeter only keeps track of 100 IP adds at a time… so if I’m still in the site, wait until 100 other IPs do somthing… then click into a new window, or even refresh, it counts me as a new visitor.

Add in frequent flyers like I am here… where when I’m in a thread I refresh every few minutes to see responses… and sitemeter becomes a horrible measure of amount of discrete visits, or visitors.

My guess is that Kos has about 1/10 of the amount of visitors as they say they have… because if there is 200 or 300 discrete visitors there at any one time, base sitemeter will see ANYTHING they do as a new visitor.

A better measure would be to just count the Opening or homepage of the site… but then you would still count people refreshing the home page as discrete visits…

Romeo13 on October 4, 2007 at 7:23 PM

The only way to prove this on our end would be to compare our SiteMeter stats to the other stat programs we use.

That wouldn’t really prove anything, because different stats packages use different metrics for “visit.” It’s not a standardized thing. This might be able to test the hypothesis: put a free SM counter and a “pro” SM counter on the same high traffic site. Check for any significant difference in the numbers the two counters report.

But there’s a gotcha there — they counters may conflict. Might be easiest to just ask SiteMeter what’s up.

Mark Jaquith on October 4, 2007 at 7:23 PM

Sounds like someone has been defrauding their advertisers. Maybe that Fitzmas guy can investigate this now that he is done charging people for crimes that never occurred.

JayHaw Phrenzie on October 4, 2007 at 7:23 PM

LOL
wha???
so, sitemeters don’t really count for anything.

bridgetown on October 4, 2007 at 7:31 PM

Whatever Kos’s actual numbers, it’s safe to say that he’s as big as Instapundit and MM combined. Which is pretty darned big.

But this includes all those Diaries, no? Between diaries and other bloggers at Kos, they have a huge staff, right? Here it’s pretty much just you and Bryan.

RightWinged on October 4, 2007 at 7:35 PM

As far as advertisers go, this is a non-issue. Kos isn’t defrauding anyone, and it makes us sound stupid to suggest it.

Anybody spending sweet cash money for ads is going to have some metric by which to analyze the cost effectiveness of the ad in whatever different media it is placed. I seriously doubt the free version of SiteMeter is the basis of that metric.

The inaccuracy of SiteMeter has been well known since before I started blogging, but the inaccuracy is global, so it serves its purpose as a comparative tool, like a bar graph with no scale– you can still tell which one is bigger.

Ruffini’s most interesting observation is that the error appears to be exponential and not linear– which is worth a “hmmm, interesting!”, but the basic truth still remains. HotAir = big. Kos = bigger.

On the other hand, it is nice to know that when the KosKids are finally stirred up enough by the Ruthuglican Theokkkracy to start the revolution, it’s going to turn out to be just a couple dozen art majors and a score of geriatric Sausalito attorneys who wear bandannas and Dead shirts on the weekends.

a4g on October 4, 2007 at 7:53 PM

Why get into the semantics? Kos = libs. Of course they are exaggerating. That’s what they do…

unamused on October 4, 2007 at 8:00 PM

“navigate away” is a bogus concept. At any given time I might have three or four different Firefox windows going, with as many as a half-dozen tabs per window. When I flip over and look at another tab, I’ve sent nothing out on the wire to tell SiteMeter about which page my eyes are on at any given time.

The only thing it could look at is the ‘referer’ [sic] to determine whether the user came to the page by clicking on an internal or external link.

The Monster on October 4, 2007 at 8:17 PM

Well, he’s mostly right. I don’t believe the 60% figure is accurate, but he’s right about SiteMeter counting visitors twice if you’re still around while the visitor list cycles through.

I hate being wrong. But I was wrong.

Meryl Yourish on October 4, 2007 at 8:48 PM

This kind of BS is why we would NEVER, EVER agree to any “pay per click” deal. We’ve experimented with a variety of counters and services and have concluded that they all count something….but what??? You can often get contradictory results out of the same service by monkeying with the setup and/or the reports.

We ended up installing our own counter to count visits to our web site our way. We can’t compare the results to anything else, but our counter does tell us which of our catalog pages people look at…and how much looking they do…when they buy our products. We can only prove that this picture works for us on our site.

landlines on October 4, 2007 at 8:50 PM

Kos should ask President Dean and Senator Lamont for more regulation of the blogosphere to make sure we all have accurate visit counts.

pedestrian on October 4, 2007 at 9:03 PM

Kos’s traffic is inflated…??

It would have to be, it’s so bloody boring, and not reader friendly at all… not that I go there often, but sheesh it’s awful.

4shoes on October 4, 2007 at 9:09 PM

On the other hand, it is nice to know that when the KosKids are finally stirred up enough by the Ruthuglican Theokkkracy to start the revolution, it’s going to turn out to be just a couple dozen art majors and a score of geriatric Sausalito attorneys who wear bandannas and Dead shirts on the weekends.

While I find the ad revenue thing interesting, your point above is the gold in this story to me.
Remember, this isn’t just a blog, it’s a movement. It attracts Presidential candidates to it’s convention. It owns the Democratic party!
So yeah, I’d be thrilled to find out that they aren’t as big as they think they are/claim to be. I don’t like their influence on politics, and I’d be happy for Democratic government officials to realize it’s ok to stop pandering to them.

MayBee on October 4, 2007 at 9:39 PM

Why get into the semantics? Kos = libs. Of course they are exaggerating. That’s what they do…

unamused on October 4, 2007 at 8:00 PMM

Kinda like “Internet inches” …

Ali-Bubba on October 5, 2007 at 12:18 AM

Wouldn’t whatever happens at DailyKos also apply to other blogs of similar stature? SiteMeter is crap, anyways. If there is to be any real check on traffic among the major blogs, they should all publish their Google Analytics stats publicly.

Seixon on October 5, 2007 at 3:28 AM

“Sounds like someone has been defrauding their advertisers.”

Well, it’s happened before. The fuss over Google’s “click through” fraud is still reverberating.

georgej on October 5, 2007 at 9:16 AM

If you look at the top 10 blogs, by traffic, 8 are “conservative”. Certainly those blogs double count as people move between them.

So… where are the companion blogs to the Daily “Screw’Em” Kos and why doesn’t their traffic reflect the same symbiotic relationship?

Further, it wasn’t so long ago a number of lefty blogs were exposed for inflating their stats. Who says such lessons went unlearned?

Finally… if 1/2 million people visit the Daily “Screw’Em” Kos daily why do so few attend their “convention”?

DANEgerus on October 5, 2007 at 10:22 AM

Zzzzzzzzzzzzz.

Huh, what?

*cough*

Zzzzzzzzzzzz.

Kensington on October 5, 2007 at 4:12 PM