Split testing ads in Google Adwords and Bing Ads

Split Testing Ads

Split testing ads is a marketing process where you pit a new design against the control. The control is your known working good version. The control version is working and you know your metrics.

The metrics can be views, and subsequent clicks or views and completions. Essentially you want to know for how many people see the piece how many take action. The Internet makes this much easier to measure than other media like radio, TV, or direct mail.

You can split test your Google/Bing/Facebook/LinkedIn Ads and you can split test your landing pages as well.

Too often people fall in love with what they like. In marketing this is huge mistake, because often what we personally like or think is clever doesn’t work. You have to test! The right answer is the one that resonates the most with your market and produces the highest conversion factor.

Below is a test of two Bing Ads. They look virtually the same.

The ONLY difference is the capital “F” on Free, and it produced a 9% increase in click through.

 

Split testing ads

Split Testing Ads produced 12% more Click Through

Little changes like this can produce big differences in the results. Below you will see an ad that pulls (gets more clicks) even better, it has a .71% click through. Again it is just a slight variation in the text, everything else remained the same.

 

Split Testing Ads

Testing means changing only one thing and seeing what changes. You eliminate the losers and continue to beat your controls.

Power Tip: Do this testing on Bing Ads, because Bing has the exact same audience as Google, and it is a lot cheaper. Test until we know we have it working well, then we can move the ad to Google where we will likely pay more for the same click.

People make the mistake of writing ONLY one ad, and therefore judge their success based upon just that one ad. Create at least ten variations of an ad when we first start out. Let those ten run for a week or two until we have enough data to make a decision about which ones are winners and which ones are losers. Turn off the losers and create some additional variations of the winner to see if we can beat it. You should do the same.