November 22, 2005

Weighing blog traffic.

Duncan Black is touchy about something I said about his traffic. I suggested he gets a lot of traffic through minimal posts, including many "open thread" announcements. That is, people use the main page as a portal to the comments window. Of course, he should be touchy, because he makes a lot of money from ads, the advertisers rely on the traffic numbers, and the ads do not appear on the comments pages.

His average daily traffic, per Site Meter, read today, is 169,709. The average visit, however, is a mere 4 seconds, perhaps as readers jump over to the comments window, which Site Meter doesn't measure. By contrast, my average visitor stays 181 seconds. So, though I lack his immensely high traffic, if you multiply my daily traffic (10,827) times the seconds spent looking at the page, you get 1,959,687. Do the same multiplication on his numbers, and you get 678,836. Isn't that interesting? I'd like to see someone do a ranking of the blogs based on this statistic! On this standard, I even beat Kos! With 2 second visits, he's at 1,485,224.

Yeah, would somebody please produce this ranking?

UPDATE: Or am I wrong in the way I understand how Site Meter counts visit length? Here's the first comment on this post:
Ann you don't understand that sitemeter only measures a visit length when you move to the comments window. If a blog had no comments then visit length would hover near zero except for those few people who clicked on the archives. His lower visit time means that there are a lot of zero length visits which stay solely on the main page. Your higher visit time means that you have many active participants in your comment section. I don't think you proved what you thought you did.


ANOTHER UPDATE: Visit time is inaccurately measured by Site Meter, but it's still something, otherwise it would make no sense for them to include it on their charts. And I hardly think people stay on Eschaton's main page, since it's full of very short posts, often simply announcing an open thread. Some portion of zeroes are apparently averaged in, where Site Meter records no movement within the blog, but we all get zeroes averaged in. I'm just pointing out an extreme difference in these numbers. Today, November 26, my average visit time is up to 3 minutes and 48 seconds. I think avertisers ought to know that some blogs keep their readers on the page much longer than others. This is an important factor, even if a fully accurate statistic can't be produced.

22 comments:

Elliott said...

Ann you don't understand that sitemeter only measures a visit length when you move to the comments window. If a blog had no comments then visit length would hover near zero except for those few people who clicked on the archives. His lower visit time means that there are a lot of zero length visits which stay solely on the main page. Your higher visit time means that you have many active participants in your comment section. I don't think you proved what you thought you did.

Ann Althouse said...

Oh, I see what you mean... I'll correct.

Anonymous said...

Your posts are long (very long). Atrios' posts are short, and primarily filled with links to other bloggers.

Atrios has lots of open threads.

Site meter counts as a visit the time between two successive page loads. This ignores the behavior of users of tabbed browsers. (This effect would be the same for both of your sites.)

Robert said...

You can't really do the analysis that way. (Well, you can - who's going to stop you? Superman won't be back until summer 2006.) It doesn't mean much, though.

The only method Sitemeter has to ascertain visit length is to compare the next hit it receives from the same IP address on another page of the same site. If a reader leaves the site, it assumes a visit length of zero - SM has no way to know where the visitor went, or when.

Say I visit www.althouseforscotus.com and spent four hours on the main page, scrolling and reading the many entries. I then type "www.hotalbanianwomen.com" in my browser window, and surf away. Sitemeter records one visit to your site, one page view, visit duration = 0.

Alternatively, I visit www.kosisatool.org, and every ten seconds I click on a new link on the same site. Each time I click within the site, the previous site gets credited with a visit duration of ten seconds (or however long it takes before I click). The last visit - the one where I leave the site to go somewhere else - will have a duration of zero.

So the page where the viewer spent four hours shows a duration of zero; the page where the viewer spent only brief periods shows a duration of around ten seconds, and looks much more "sticky". But in fact the opposite is true.

Short, non-zero site visit durations (like four seconds) most likely indicate someone coming to the site, and clicking on story threads to open them in new windows or tabs. Long visits generally indicate someone coming to a site, reading a page for a while, and then clicking over to another page on the same site.

In other words, it's likely that your visitors and Kos' visitors have a different usage pattern owing to the variation between your site designs, not that the Kossacks only can stomach reading the page for four seconds before they have to flee in disgust.

That job is left for the rest of us.

Ann Althouse said...

Does Site Meter ignore the zeroes when it does the average visit length? The zeroes, as you're explaining it, represent the failure to measure the visit length. I suppose clicking into a Haloscan comments window would be a zero, the equivalent of going to another website.

vbspurs said...

Tangential aside:

It's no mystery why Atrios has a lot of blog traffic -- and that reason is an open-forum based blog.

I've mentioned this point a dozen times in various guises, but takey-look see here, how these blogs lean.

Higher Beings
1.Instapundit.com - Right
2.Michelle Malkin - Right
3.Daily Kos: State of the Nation - Left
4.Power Line - Right
5.Wedding Cakes - Cakes?
6.Stop The ACLU - Right
7.lgf: those who are tardy do not get fruit cup - Right
8.Captain's Quarters - Right
9.Boing Boing: A Directory of Wonderful Things - Non-political
10.Hugh Hewitt - Right

You have to go further down to number 12 to get a left-leaning blog, Atrios' own.

Mortal Humans
11.Mudville Gazette - Right
12.Eschaton - Left
13.DRUDGE REPORT 2005® - Right
14.The Volokh Conspiracy - Right
15.Wizbang - Right

The fact of the matter is, most of us have a cornucopia of choice of elite right-wing or libertarian blogs.

People in the progressive side do not.

Instead of being touchy, Atrios should be happy he's got the traffic he does, lasting whatever time it does, because if there were more like him out there, the gruel would be thinner for him.

Cheers,
Victoria

Robert said...

Ann, I believe that SM does include the zeros in its average (and thus systemically understates every site's visit duration somewhat). Your 181 seconds becomes all the more impressive; Kos' 4 seconds somewhat less damning.

vbspurs said...

I went to Site Metre's Help area:

>Why do some of my visitors have visit lengths of 0:00?

That means the visitors are only staying to view a single page and then leaving. The only way that Site Meter knows how long someone is on a site is by the times of each page view. If they only look at a single page and then leave, we don't know how long they looked at the page. If they looked at two pages and left we would know they at least were on the site during the time of the first page view and the second page view. The difference between those two times would be the length of the visit.

And:

What is the difference between a visit and a page view?

When you are browsing a site, every time you follow a link to a new web page, it is treated as a single page view . Site Meter defines a visit as a series of page views by one person with no more than 30 minutes in between page views. If you click on a link to another site, and then come back to your site within 30 minutes, you are still on the same visit and Site Meter won't increment the counter. But Site Meter will increment the number of page views recorded for your current visit.

So what Robert stated about impressiveness stands.

Cheers,
Victoria (2:36 avg. on Sundries)

reader_iam said...

This has truly been useful and informative and has answered a number of questions I had.

Thanks.

dave said...

Or am I wrong in the way I understand how Site Meter counts visit length?

Oh, heaven forbid you should then not make blanket claims... or even worse, take down your erroneous post.

Enjoy the spike in your traffic while it lasts, dearie... but realize people are only dropping by to laugh at you.

Ann Þø said...

I read your blog because one third of the posts are ones I want to read (on average).

AND I LOVE YOUR COMMENTERS.

This is what (to me ) is the real added value.

One day I will give you my ranking of your commenters.

Keep the good work

Ann Althouse said...

Dave: My visit length has been around 3 minutes for a long, long time. It has nothing to do with the current controversises. But I can see how you might bumble into such a notion.

vbspurs said...

Enjoy the spike in your traffic while it lasts, dearie... but realize people are only dropping by to laugh at you.

Yeah, because it's plain as plain as your nose to see 3,000,000 visits and counting makes Ann a Johnny-come-lately to Blogosphere.

Grow up, you pus-filled, crotchety troll.

Anonymous said...

That's so cute! Ann has a groupie!

Howard said...

A couple of observations.

It's possible (I won't tell you how) for outsiders to use Site Meter to gain entrance to your blog, at least it does on Blogspot. I was going crazy with trying to find out how people were nightraiding me. I finally removed the Site Meter logo from my site.

Number of hits or readers or what: I spent the better part of a year trying to figure out why my email count was so high, comments so low, and supposed readership was so low. Then all of a sudden, like out of dam near nowhere my readership jumped from 5K per week to the current 35K per week IN LITERALLY A SINGLE DAY. How can that happen? I changed nothing. I've verified the hits and they're accurate.

I suspect there may be some glitches in the Site Meter for some of us

Razib Khan said...

victoria, we're talking about traffic, why are you linking to the ecosystem linking rankings. here is the traffic ranking top 12:

1) Daily Kos: State of the Nation 742612 visits/day (3)
2) Gizmodo, The Gadget Blog 297682 visits/day (161)
3) Eschaton 169709 visits/day (11)
4) Gawker, Manhattan Media News and Gossip 165307 visits/day (157)
5) Instapundit.com 156197 visits/day (1)
6) Defamer, the L.A. Gossip Rag 129460 visits/day (403)
7) Common Dreams | News & Views 105774 visits/day (52)
8) A Socialite's Life 102896 visits/day (1473)
9) Crooks and Liars 102329 visits/day (39)
10) Michelle Malkin 100993 visits/day (2)
11) lgf: those who are tardy do not get fruit cup 95369 visits/day (6)
12) Pink Is The New Blog | Fingers Firmly On The Pulse 83460 visits/day (1500)
13) Autoblog - www.autoblog.com _ 83356 visits/day (1934)
14) Go Fug Yourself 82707 visits/day (270)
15) Power Line 82064 visits/day (5)

your presentation of the data elided over the reality that daily kos is the 700 lb gorilla of the blogosphere in terms of traffic, having long since surpassed atrios and instapundit.

whether ann is right about atrios' quality of visits, i think it is safe to say that the quality of visitors to high traffic weblogs is pretty low. whenever my weblog (~3,000 uniques/day according to sitemeter for a few years now) get's a link from the corner or (long, long in the past) instapundit el stupido's starting swarming. seeing as 98% of the population is dumber than 130 IQ, i guess it makes sense you'd start having to sample the whole spectrum if you went north of 10,000 uniques (sorry ann :)

Ann Althouse said...

Howard: What are you talking about? I find what you're saying scary, but incomprehensible.

Nick said...

You're also forgetting about all the people who read blogs through RSS feeds... which aren't counted by Sitemeter at all.

Even ATOM feeds which are restricted to summaries, like yours aren't perfect. Maybe you've noticed (I Have) that Blogger has a bug where it often times still puts the full post in the ATOM feed even when you have it set to summary only.

AnechoicRoom said...

Well now, we've learned something. Ain' dat cool when it happens.

A little while back over at DtP, before the coup de grace, there was some/a little straightforward blogging about eyeballs/advertising.

Den Beste opined, it wasn't the number (of eyeballs), but what was behind them.

No comment or inference on what is behind Kos Kiddy eyeballs (or lack thereof).

kc said...

Grow up, you pus-filled, crotchety troll.

I shall cite this comment as an exemplar of the rampant anti-male sexism on this site.

Finn Alexander Kristiansen said...

Kc:

Dave said...
Enjoy the spike in your traffic while it lasts, dearie... but realize people are only dropping by to laugh at you.

Vbspurs then made her, uhm, "anti-male sexist" (in your words) comment in response to that post.

Kc, I would think that if a person drops onto a website, and posts something relatively stupid or inaccurate, that individual should at least expect something verbally pungent in return.

If this site is indeed sexist against men, and frankly that's news to me, then I must be an accidental masochist, and you, KC, must be smarter than readily apparent, in some mystical, pull a thought out of your bum kind of way.

I assume you were speaking in jest.

Anonymous said...

I love Ann's blog because it's like a zoo where we can see endangered species, like Ann and Victoria, in their natural habitat of ignorance and delusion.