When selling your ad inventory via a CPC network you don’t just need high bids, you need clicks too. Imagine the extreme situation where an advertiser bids very high on an ad that users of your site will never click on. That bid will win often win the auction for impressions, but will generate no revenue if no user clicks on it. The result can be supressed earnings.
Whilst systems such as the quality scoring of adverts ensure that inventory isn’t dominated by high-bidding un-engaging ads, these are less able to cope on a per placement basis. Imagine the situation where a website about electric cars is attracting high-paying ads for the latest gas-guzzling 4×4. That ad might perform brilliantly across the network as a whole, but poorly on a site where the audience is opposed to the content.
Many site owners try to combat this by blocking categories of ads that they feel are likely to perform badly. Get it right and this can be a useful tactic, but get it wrong and your average CPC can tumble as price pressure is removed from the auction.
The best answer is often to use your understanding of your audience to identify opportunities, but split test the results to measure the impact.
Split test blocked categories in AdSense
In-account experiments provide a very easy way to run these tests in AdSense, even removing the need to change any code on your website. Once logged into AdSense navigate to My ads > Experiments and click the “create” button under “Allow & block experiments”.
From there you can create an experiment that blocks the categories that you think might supress earnings. Google will run the experiment for you showing the original ad choices to some of your audience and the limited list to others. As the experiment runs it will measure the relative earnings of each approach and report back to you when it is confident in a result.
Split test blocked categories in on other networks
Publishers using AdX or another ad network will most likely need a third party script of some sort to handle the splitting of traffic. If you already use an Ad-Server, such as Doubleclick For Publishers, then this is ideal.
Simple create two ad tags in AdX: One for the original set-up and one that has your chosen categories blocked using rules. Traffic a line item into DFP that uses each of those tags and have them compete on equal basis for traffic. You can, for instance, give them both a fixed share of impressions. Checking the RPM of each tag will show you which set-up performs best.
A word of caution
The ad auction can sometimes be a delicate structure. Removing bidders can sometimes have a big effect and the more advertisers you block the larger that impact is likely to be. Always proceed with caution blocking categories and always measure the results.
