Chitika is a large advertising network with a lot of sites in their portfolio. For a sense of the scale of numbers they’ve done this analysis on, the sites they included in this analysis racked up a total of over eight million impressions in May this year, and May’s not even over yet! So let’s not misunderstand, this is a lot of sites with a lot of traffic in this study.

Their findings confirm something we’ve believed for a long time - the first spot is worth a lot more than the second spot. A lot.

Results in number one ranking on Google achieved nearly 35% of all the Google traffic on the Chitika network in this time frame. That’s more than double the traffic that came from second place results.

But this isn’t quite as clear cut as it seems, despite the huge numbers involved. Bigger samples don’t make for better research. I have a problem with the way this is presented, because Chitika don’t mention the inherent biases in their article.

Biased data

This is not a controlled test. To make this a perfect experiment, there would be one site, and 10% of searchers would see the site in position 1, 10% in position 2, etc down to position 10. That way you would know that you’re not being affected by other factors. But that can’t happen. Instead different sites have ranked differently for different keywords. The biggest factor at play here is the size and recognition of the sites. It’s not unreasonable to suppose that the sites that rank well are the bigger, more well known sites in their fields. In which case people are more likely to click on those sites simply because they recognise the URL. That would imply a bias that sites higher up the rankings would get more clicks.

Secondly, we don’t know if brand terms have been included in this data. If you searched for Periscopix we would likely be the top result. And we’d probably be the only link on the results page that gets your click. The bias there is enormous, once again skewing the data to make it look like top position is worth more. But really, top position just means it might have been a brand search.

I want to believe Chitika’s results, because they seem very natural to understand and they support all the anecdotal data we see, and knowledge about they way we search ourselves. But the fact they haven’t mentioned these issues that would make the data look much stronger than it really is does undermine their cause a little.

Share this article

• Tweet

• How Google’s auction dynamics affect your pay per click bidding strategies

  17th April 2009