Hits vs Comments

Blogging controversy can generate a lot of comments, but it is often said that this also generates a lot of “hits” (page views) and sometime we bloggers are accused by people (often commenters) who don’t blog themselves and don’t have access to the secret data about blog sites that bloggers can see of generating controversy in order to generate hits.

Stephanie Zvan has addressed this issue, just now, in a post on her site. A highly controversial post that is generating a lot of comments, but also, a post that has a video of a cute little kitten. If that post generates a lot of hits, it will be because of the kitten, not the controversy.

For my part, I went and looked at the seven posts with the most page views from a recent month on one of the blogs I write. The data are here:

i-99528063127d7e746e94e70a19e383a5-hits_vs_comments.jpg

Notice the R-squared value. The X axis is number of hits, the Y number of comments. Clearly, there is not much of an effect.

Someday I’ll do a more intensive study, one with a larger sample size, and thus, probably, a less controversial result.

Comments

  1. #1 Ole Phat Stu
    January 6, 2012

    Bet that single point on the right is an outlier.
    What would slope and R be then?

    Interesting that you’d do a controversial post about controversy, sort of recursive ;-)

  2. #2 Doug Alder
    January 6, 2012

    One of the problems with counting hits is separating out the bots from the eyeballs. It’s not easy and takes a lot of work. If I look at one stats program in which I have no filters running I can see 300-500 hits per day whereas in a different stats program where I have blocked known bots from registering the eyeballs count is only 10% of that total. The vast majority of traffic to any website are bots of one sort or another – advertisers who are paying x cents (or dollars) per impression are being ripped off, big time.

  3. #3 Greg Laden
    January 6, 2012

    Doug, I’m pretty sure that applies mainly (oversimplifying a bit) to the first few dozen or hundred hits. A web site with virtually no content or activity will be “visited” by these robots. But a web site with thousands of daily actual human visitors will have a sufficiently high ratio of real to robotic that variation in the numbers is mainly accounted for by humans (though I’m sure there are also more robots)

  4. #4 Mike
    January 6, 2012

    Interesting, Greg, and consistent with my experience. Then again, 10,000 Birds rarely incites controversy. When we do, we generate an exceptional number of comments but rarely a noticeable increase in traffic.

    The key to comments, IMHO, is community. When a blog becomes a hangout, readers start to use comment threads to converse back and forth. I’m sure you see this all the time in different nodes of your bloggy empire.

  5. #5 Doug Alder
    January 7, 2012

    Greg – there’s that , and for sure it’s significant, but there’s also a large increase in bots for all the cross links. You linked to Stephanie’s post – guarantee you the bots scouring your post followed that link over to hers. Posts that generate a lot of controversy end up being linked to from a great many more sites which in turn causes bot traffic to spike. i realize I can’t really use my site as a good example since I have very little traffic – but it’s the only one I have the stats for so it will have to do :)

  6. #6 Greg Laden
    January 7, 2012

    OK, fine, but what is your basis for saying his? Do you just believe in Bots and everywhere you look there is a bot hiding? Or is there some study of bot behavior and enumeration one can refer to? How do I know YOU are not a BOT???

    Anyway, as I understand it, Statview, Google Analytics, etc. all have facilities for identifying bots and removing them from the count. These methods are imperfect, and that imp refection may account for much of the difference between different counting methods. But in the end, page views counted on your server log woudl include all the bots, and page view counting services would remove variable numbers of, but hopefully a good number of, bots.

    The key is this: Variation among pages on a blog or blog network is accounted for primarily by user behavior. You’ll notice (and it’s funny that no one has mentioned this) that I’ve trimmed out the numbers on the graph above. I’m focusing on variation not numbers party becuase I consider the numbers to be proprietary information and partly because they are not real in absolute terms (because of bots and other sampling issues) but I accept the variation as meaningful.