A Longitudinal Study of Blogging Traffic

Back in March, I noted that I had inadvertently done an experiment to see what kinds of posts bring the most hits. That week, I posted one peer-reviewed post every day, along with a bunch of other articles, and I looked at the traffic stats to back up my contention that hard-core science blogging is not what racks up the page views.

The question came up again at the conference this past week, which reminded me that at the time, some people argued that the science posts weren’t a big immediate draw, but would build up more posts over the long term. I thought it would be interesting to go back and look at the data in Analytics, and I generated the following table comparing page views from the week of March 9-15 to the page views from March 9 to September 12:

URL Mar. Views Sept. Views Change Mar. Rank Sept. Rank Rank Change
Talk Like a Physicist 1215 4510 3295 1 1 0
Pimp Me New Blogs 453 524 71 2 5 -3
A User’s Guide to Vacuum Pumps 432 864 432 3 3 0
Tips for Speakers 333 367 34 4 7 -3
A User’s Guide to Vacuum Pumps (part 2) 331 523 192 5 6 -1
FutureBaby Playlist A-B 317 333 16 6 8 -2
Art and Animals 275 276 1 7 14 -7
Lab Visit Report: Biophysics 258 293 35 8 12 -4
Lab Visit Report: Francium 256 730 474 9 4 5
Lab Visit Report: Cavity QED 251 322 71 10 9 1
Lorentz Contracted Asteroids 244 258 14 11 15 -4
Nice Beaver 243 879 636 12 2 10
Lab Visit Report: Cold Plasmas 236 282 46 13 13 0
Headline Mismatch 193 198 5 14 17 -3
Score One for Physics 184 181 -3 15 19 -4
Lab Visit Report: Four-Wave Mixing 179 305 126 16 10 6
FutureBaby Playlist C-G 174 174 0 17 20 -3
Interstellar Economics 169 257 88 18 16 2
Imperturbable Ducks 146 297 151 19 11 8
Sloth-in-a-Box 139 186 47 20 18 2

So, what can we say about this?

Well, I’d say it’s a mixed bag. There are science-related posts that saw big gains between March and September, but the proportion of science content in the top ten posts doesn’t change–in fact, the second most popular post over the full span is just a picture of a zoo animal with a movie-reference title. One research blogging post moved into the top ten, but it just replaced one post that fell out– the total number of serious science blogging posts in the top ten remains the same.

It is true that if you just look at the top ten posts in terms of the change in the number of views, five of them are science-related (three research blogging posts and the two vacuum pump posts), but the other five are fluff of one sort or another.

This is a small sample of posts, of course, and there are innumerable factors that can influence these things, but at the very least, I would stick to my original conclusion that serious, detailed blogging about science is not a significant driver of traffic for this blog.

Which is not to say that science per se does badly– life-in-the-lab stuff does quite well, as shown by the performance of the vacuum pump posts. I have no intention of stopping that stuff, and as I said in Waterloo, I think it makes a useful contribution to the community. But it’s not what brings in the big bucks.

8 thoughts on “A Longitudinal Study of Blogging Traffic

  1. I find myself thinking along the same lines. Sometimes I go for the cheap score and write something that I know will be popular. The other stuff is still useful. My thought is that if it is “out there on the internet” someone might find it useful someday (probably after I die).

    Nonetheless, good article.

  2. Compare physics and chemistry crackpots. What can ignorance (educable) or stupidity (permanent) spew about the Sonogashira coupling or NMR? Nothing! Physics is the weaker discipline by public declamation.

    Chemistry need only make Spandex, scrubbing bubbles, and Cialis but physics must be rendered cute to be attractive. Pimp your tensors! Won’t you all look a right proper Charlie if the vacuum diastereotopically interacts with enantiomorphic mass distributions? Chemists do it all the time, especially illegal pharm-workers.

    (“What is purple and commutes? An Abelian grape.” No contest.)

  3. You will and Scienceblogs will eventually lose readers if there were no science in the blogs. Or rather minimal science. The noise ratio is getting pretty high now with all the election stuff. Hopefully a few month phenomenon.

  4. You will and Scienceblogs will eventually lose readers if there were no science in the blogs. Or rather minimal science.

    I doubt it.
    Or, rather, we’ll lose some readers, and some types of readers, but gain a different sort. All that will happen is that the character of the audience will change.

    I’ve seen several blogs shift from posting mostly science content to mostly political ranting, and from what I can tell, their traffic has only increased.

    The noise ratio is getting pretty high now with all the election stuff. Hopefully a few month phenomenon.

    There’s a new group blog specifically about the election starting up soon, which may help to localize the political content a bit.

    I’ll probably have a few more politics posts than usual in the next few weeks, but I have no plans to shift to a politics-heavy format for the simple reason that when I spend a lot of time writing about politics, I start to turn into a tedious asshole, and I don’t like that.

    Of course, I’m probably going to bore the pants off of my usual audience with a whole bunch of “Web 2.0” stuff as I write up stuff from the conference last week, instead…

  5. There is (at least) one obvious anomaly in the chart: The number of page views in a given period should be positive semidefinite, but your “Score One for Physics” post seems to have had -3 page views for the period 16 March-12 September.

    Here’s a WAG for why some posts get more page views: If there is an active discussion thread attached to a post, I’m much more likely to return to that post, especially if I have commented on that post myself (in part because one of the subsequent commenters may have responded to a point I raised). Caveat: I haven’t taken the time to figure out how to do RSS (which is probably just as well as I would end up spending way too much time on blogs). I don’t know what fraction of your audience reads blogs the way I do.

  6. There is (at least) one obvious anomaly in the chart: The number of page views in a given period should be positive semidefinite, but your “Score One for Physics” post seems to have had -3 page views for the period 16 March-12 September

    Yeah, I forgot to mention that. Google Analytics occasionally does some weird things.

    I trust the numbers to give the right general trend, but when you’re talking about small number of hits, I think it’s pretty worthless.

    I haven’t taken the time to figure out how to do RSS (which is probably just as well as I would end up spending way too much time on blogs). I don’t know what fraction of your audience reads blogs the way I do.

    I have a bit more than 1,800 subscribers to the feed for this blog, and occasional posts end up in other feeds supplied by SB. I get about 2,000 page views on a fairly typical day, which I don’t think includes the RSS readers, unless they click through to see a full post.

  7. Cool. I think you’re right about it all being about the character of the audience. It’s a shame there’s not an easy way of filtering out non-readers: people who arrive from Google Images, stay on the site for a second and then leave, for example.

  8. I think we shouldn’t pay attention so much to hits. It’s an imprecise and irrelevant measure. As Euan said above, a large part of the people who visit come by accident, are not interested and are not actually readers anyway. The chance that one of them becomes a returning visitor is negligible. Did you ever add a blog to your feed that came up in a keyword search but you saw at first sight it didn’t answer your question? To give you an example: my visitor count increased dramatically after I moved to Waterloo. Why? Because the site where I wrote about the way my furniture took became the first hit on Google images for “map of America” (despite the fact that it didn’t even have a map of America). And who cares about that?

    Do you recall that in my opening talk (sorry, not-an-opening talk), I brought up the question: What feedback improves quality? Well, that’s the problem here, you’re worried about a feedback that would lead you to optimize performance according to a measure that’s irrelevant for the scientific content. Increasing the numbers of visitors doesn’t improve neither quality of writing nor accuracy. I’ve been offered to join scienceblogs a while ago, so I know you get money by hits. But what kind of incentives does that set? Shouldn’t Seed be more interested in high quality than in many hits? Who makes such decisions and what was the thought behind that? (Yeah, I know where the problem comes from, I only have to look at the ads in your sidebar.) Bottomline is, your blog is great, stop counting visits.

Comments are closed.