What is A/B Testing? I’ll Tell Ya
by Allison Otting • February 24, 2014
A/B Testing isn’t quite as bloody, but it’s definitely a battle to the digital death. (image source)
A/B Testing is exactly what it sounds like. You take item “A” and you test it against item “B.” You then gather the data, and you are able to tell who is the victor.
In pay-per-click marketing we use A/B testing to optimize our campaigns and make sure that we have all the best performing elements running rather than a bunch of duds. So, what is A/B testing? It’s what makes PPC Marketing less of guesswork and more of a science.
But how does it work in PPC?
Well, let’s take a look the elements of a paid search campaign. First, we have the keyword.
This is what someone is going to type into Google, in hopes of finding a frozen banana. Maybe it will be dipped in chocolate! Maybe it will have sprinkles!
Whatever they are looking for, if you’re in the frozen banana business you would want to get this person to buy your frozen bananas. Every A/B testing journey begins with this: a business goal. Every test after that goal is established is a step towards a better banana business model (sorry, I couldn’t resist).
You want that keyword to lead to your business, so you’re going to write an ad and bid on the keyword “frozen bananas.” You put it into place, and it seems like it’s doing okay, but you want to know for sure. This is when we start our A/B test! You need to make a second ad.
What data qualifies an ad as a winner? Click-through rate! You will have a certain amount of “impressions” or views of the ad, and then you’ll have a certain amount of clicks on the ad. You will divide your clicks by impressions, and then you have your click-through rate (or CTR, as most marketers will call it).
100 clicks/1000 impressions = 0.1 or 10% CTR
Now you have your ad winner, but where are these clicks taking the user? Does the ad take them to a homepage? If so, you are throwing away so many potential clients! There is a lot of information as to why you need a landing page, but for our purposes we’re going to focus on the fact that you can also A/B test landing pages!
The goal of landing pages is for the customer to convert, whether that means a phone call or filling out a lead-gen form. By taking one of these actions, you now have the means to contact and sell your product or offering to that person? Pretty cool huh?
You want to make sure that your landing page is conversion rate optimized, and while there are guides to doing so, things change depending on your offer and your customer base. So, you need to make at least two landing page variants and have them duke it out! After a certain point, just like the ads, you’ll know which one will convert better.
Make sure that you don’t change too many things on a variant. If nothing is the same, you don’t know what’s helping or hindering the landing page. The best practice is to change one thing at a time whether that’s the headline, call-to-action, hero shot, etc. By changing one thing at a time, you’ll know exactly what’s causing any change in your conversion rate.
Feeling like an A/B tester champ already? Tell us in the comments below what best practices you have found in A/B testing.