A/B Ad Testing

Ad testing—it’s something we do every day, and it recently struck us that we’ve never written about it.

Something we have written about several times before, though, is Conversion Rate Optimization (CRO). As a refresher, CRO involves examining every decision point in the customer journey and exploring ways to improve the outcomes at each of these points. Ad testing is a perfect example of this.

A/B Testing

If you remember anything from middle school science (other than looking at grass under a microscope), you may recall that one of the essential elements of a good experiment is to only have one variable between your control group and your experimental group.

If you give one fern more light and play classical music for it, you’ll never know if it was the music or the sunshine that caused it to grow larger than the other fern. You have to isolate the variable, and that’s what A/B ad testing is all about.


Here’s where it gets fun (if you’re a person who has fun easily). What piece of the ad is going to be your variable? The headline? The call-to-action in the description? The landing page URL? Choose one, decide what the “B” version will be, and leave everything else the same. Let both ads run simultaneously for a while without touching them, and let the data roll in!

In order to successfully A/B test a pair of ads, you need to do it in an ad group that gets a significant amount of traffic. The sample size needed to make confident decisions will likely vary from account to account, so use your best judgment and err on the side of being patient so you can feel satisfied that the data is trustworthy.

Once you have enough data and can declare a winner between your ads, the options branch out from there. Are you going to take your winning ad and copy it to other ad groups? Create another variant to test against your current victor? Keep that ad running and set up two different ads to start a completely new test?

PPC Management: What to Do When the Numbers Don't Add Up

I should also add that, even though you only have one variable, you might end up with an interesting mix of outcomes on your hands. Your “A” ad may have a higher click-through rate and cheaper cost-per-click, but a lower conversion rate. What do you do then? Well, that is why you get paid the big bucks.

Or if you’d like, it can be why you pay us the (not-so) big bucks. We’ve worked on hundreds of PPC accounts over the past decade, and optimizing them to run at peak performance is our idea of fun. If you’d like to learn more about our PPC management services, shoot us a message and we’d be glad to talk!