Advertising test results.

Advertising test results.

bart's picture

We've been running test advertisements for a little over a month in order to collect advertising statistics, and I thought I'd share some highlights from the results, as a matter of interest.  First, some general notes:

A number of people suggested on the forums that we advertise with Project Wonderful.  While they seem like a great project, Project Wonderful wouldn't be a good fit for OGA, because they only run a single ad over the course of a week.  Most sites that advertise with Project Wonderful are webcomics, and most visits to webcomic sites are a single hit, after which the user leaves (that is, they read the comic and go).  What's important to note here is that unique ad impressions count for a lot more than ad impressions in general -- if someone sees the same advertisement five times in a row, they're probably only going to click on it once, and even if they happen to click on it more than one time, it doesn't really do the advertiser a lot of good.  Where a webcomic's average page loads per visit is probably just over 1, OGA's average number of page loads per visit is 12.5.  That is, when users look at OGA, they tend to stick around for a while, so we'd be a lot more effective at advertising if we show people multiple advertisements.

So, over the course of our advertising test, we ran a total of 19 test advertisements.  At one point, due to a performance issue in the advertisement module (which I eventually fixed), ads were loading 6 to 10 seconds after the page loaded, so during that period of time it's safe to say that our numbers were off.  I fixed the issue about a week ago, so the numbers from this week are a lot better.

OGA collects web statistics with log analysis, which means that we don't depend on people not blocking the various web tracking services.  OGA users tend to be more tech savvy than average, and tech savvy users tend to be more likely to block web tracking, which means our internal log statistics are a lot more accurate than, say, google analytics would be.

So now, the numbers.  Note that these numbers are not a guarantee of ad performance, they're just what we saw in the past month.

  • Typical click through rate on unique impressions was between 1 and 2 percent.
  • The consistently best performing ad was for Cube Trains, which got about a 3% unique click through rate.  I'll include a picutre of the ad below so that people can see it.
  • The lowest click through rates were around 0.4%.
  • The average number of unique daily impressions was about 475.  My aim is to sell enough ad slots to keep this number around 400.
  • Ad performance was significantly better during the first week of the ad run, with some click thru rates over 5%.  Unsurprisingly, these numbers didn't hold.  However, you can expect ads to perform a bit better than average during the first week or two.

Other interesting tidbits:

  • The ad for Angry Turtle Jewelry, which I wouldn't think of as having a huge overlap with our userbase, did about average.
  • Clint Bellanger told me that the ad clicks for Flare tended to be higher quality than random hits on his Flare site -- that is, people coming from OGA tended to stick around longer and view more pages than average users.

Here's the aforementioned Cube Trains ad:

Want to advertise with us?  Check out the Advertise with OGA page, which includes detailed pricing information and instructions for purchasing an ad slot.

Thanks!

Bart