The Sweetness of A/B Testing – With Ice Cream Examples
by Adam Kaiser • November 16, 2014
Beakers, test tubes, and smelly fumes. What’s your A/B testing cooking?
I would like to start out by asking a very important question, have you ever had to make a decision? Ok that was a trick question, unless your answer was no and if that’s the case, something may be wrong.
Deciding what’s best for you usually breaks down to, within reason, trying multiple things.
How do you know what ice cream flavor is your favorite if you haven’t tried all of the flavors? How do you know if you like your coffee hot or iced, if you haven’t tried it both ways. How do you know if you like PlayStation or Xbox if you haven’t played both? iPhone or Android?
[Tweet “How do you know which ice cream flavor is best? Did you A/B test the other flavors?”]
I could go on and on… My point here is how do you know your PPC campaigns are running as efficient as possible without a/b testing? The answer? YOU DON’T.
Ok, so you have a really sexy landing page, its conversion rate is sky high and you’re thinking… I’m such a boss.
Could it be better? Maybe, maybe not. But how will you know unless you test it? Create a split test and let the numbers decide.
Take your emotions out of the design and what you think looks and works best. I guarantee you that the majority of your visitors will give you insight as to what works best for them. And when I say what works best for them, I’m talking about what makes them take their wallets out and give you money.
It could be as simple as changing the headline, call to action, or even just the button color on your landing page.
You or the agency you work with spends a ton of time within the AdWords interface, when in reality, that’s only half the battle. Traffic is only good traffic if it converts, and the conversion happens on the landing page or on your site, not in the AdWords interface (unless solely rely mobile click-to-call extensions).
Now here is where things get really wild. Let’s say the landing page you have gets a ton of traffic and you have created your new test with a cool new headline. Why not make another variant to test at the same time? Its simple math people, according to Rhett Horton over at TestingTheory.com, your odds for finding the best option will increase when you add more experiences.
This means that you can optimize your PPC performance by finding out what works best, faster. Granted that you have enough traffic to generate fast and meaningful results from your a/b testing. You can also find out what works best quicker by testing more variables simultaneously.
[Tweet “The more a/b testing experiences you use, the faster you learn and make more profit”]
Let me break down the math for you. You like chocolate ice cream, but you want to know if it’s your favorite. So you taste vanilla and compare it to your sweet chocolate. Your odds of finding your favorite between the two are 50%.
Now let’s throw strawberry into the mix as well, your odds just increased and your chance of finding your favorite flavor just went up to 66%… but wait, there is another flavor, well try that one too, your odds have just gone up to 75%.
Now, instead of going back to the ice cream parlor every day to compare one new flavor to your previous favorite, you have just eliminated three flavors in one sitting. Same thing goes with landing pages. If you have 3 ideas, why not test them all at once and give 25% of the traffic to each variable. 75% to the experiments and 25% to the control.
So even if chocolate is still your favorite, you have figured this out in a third of the time, compared to going back and testing one flavor per day (which is something I still do, my wife just doesn’t know it).
The best part about this is that every campaign, everyone and everything is different. Even if you think you know what’s best, you really have no idea until you test it.
Every single aspect of your marketing efforts can be tested. Don’t let a high a performing ad sit there and perform well; test another, test 2, test 3? Maybe one will perform better, or maybe the original will continue to be the champ.
Either way, you’re always working to find the best option and trust me, you’ll really like it when you find the best flav…I mean option.
What have you found to be the biggest gains or hurdles in your A/B testing? Let me know!