Anatomy of a Spam Attack
Yesterday, I did some browsing of the web analytics for this blog to get comparison numbers for the browser stats I had for Connect.
Today, I went back to do a little more digging and some navel-gazing
We use Google Analytics, which I prefer to Mint for web metrics. It has loads of metrics beyond the standard pageviews and visits. As a side note, now that Feedburner accounts are merging with Google accounts, I’m hoping that Analytics will soon include Feedburner stats too. Seems logical.
Anyway, I like to set the date range to the life of this blog (from June 2007) to get the best snapshot view from the graphs.
What jumped out was the Bounce Rate graph.

All of a sudden, our normal 75% bounce rate (I know, terrible) inexplicably dropped to less than 40% a week ago and sustained that rate all last week.
Definitely weird. Maybe after the Batman vs. Superman post, everyone was extra relieved to get back the normal, hard-hitting content we serve everyday. I laughed all the way through that sentence, obviously untrue.
I relish a data anomaly, as a recovering economist, especially if there are graphs to show the patterns. I am an unabashed data pr0n dork.
Accompanying the drop in bounce rate, there were, not surprisingly, corresponding jumps in pages per visit and pageviews over the same time period. Makes sense, the longer people stay on your site, the more pages they are likely to view.

Logically, you would also expect to see a rise in time spent on the site, as people read more. Not so much. In fact, Saturday’s average time on site was 19 seconds; that same day, the bounce rate dropped to 36% from 69% and pages per visit jumped to 2.41 from 1.64.

All this points to comment spammers.
Exhibit A: Looking through the WordPress and Disqus comment logs from the last week, there was definitely a rise in comments on old posts, definitely a sign of spam. And these aren’t old posts that come up on the first page for common keyword searches, like “oracle iphone”.
Exhibit B: The spam comments are borderline, with plausible names and comments, not the usual link spam left by Monster Truck Rally. This tells me spammers are modifying their behavior slightly to get past the measures Disqus has taken.
Exhibit C: The pattern of multiple comments onm different posts from the same account backed up the web metric data.
So, I accuse Colonel Mustard, in the Study, with the lasso.
I know, as a naive kid, I thought that was a lasso. Ah, innocence.
Comment spamming has been on the rise this year, at least the spam that gets past spam filters. Disqus noted that the recent rash of spam comes from real people, not bots. The assumed goal of comment spam is to bump SEO for the spammers; I firmly believe this is a new cottage industry, operated Mechanical Turk style.
A crappy economy opens up a larger pool of people with computers who are motivated to earn easy money, and how much easier does it get than comment spam? Find a blog that allows unverified or anonymous comments and drop three comments on three posts. In and out in a matter of seconds. They probably get paid for the gross number of comments with the spammer’s link.
This might even be that job advertised on the TV. You know the one that says you can make thousands in a week, tens of thousands in a month, working “on the Internet” from home. All those smiling people tell you nothing about what the job entails. There’s always a shady URL that tells you nothing about the company.
Anyway, I’m not really bothered by comment spam, but I know people are, e.g. Bex, who uses a comment captcha process that makes me want to cry it’s so frustrating.
Does it bother you? What do think of my analysis? Did you enjoy the web analytics primer?
Sound off in the comments with something useful, like “I will give it a try for sure !”.
Possibly Related Posts
- Who Benefits from Blog Comment Spam?
- Discussing Disqus
- Leave a Comment using Facebook Connect
- Now We’re on FriendFeed
- All Your Comments Are Belong to Us
-
Sarah Smith
-
Jake
-
David Dorf
-
Jake
-
bex
-
Jake



