(713) 568-2763

Creating a Better Site Through Split Testing

Creating a Better Site Through Split Testing

Split (or A/B) Testing

Welcome to the internet! Where you can, and should be, testing everything!

In this post we’re going to be covering split testing, also known as A/B testing.

This powerful tool can be used to ensure your website is constantly and consistently improving conversions, lowering bounce rate, or working towards whatever goal you have set for it.

The general idea around split testing is that you take your web traffic and send it to more than one variation of your site and monitor the results. The better performing variations get the majority of the traffic and new variations get a smaller piece of the traffic until they are proven to outperform the current standard.

Through this type of controlled experiment, you can see measurable results every time you make a change. Instead of simply hoping that users will prefer a new page, you can know definitively what variation is better.

How to Use Split Testing

Let’s run through a few examples to better illustrate how split testing can improve your bottom line. Bob’s Brewery is bottling root beer and selling six pack bottles online. The going rate for a six pack is $3. Bob decides he wants to do a split test and see if he can sell a six pack for $9! He sets up his website to do a fifty-fifty split of traffic.

Half of his visitors get the $3 price, we will call this Variation A. The other half of his visitors get the $9 price, we will call this Variation B. During this test the conversion rate on Variation A is 30% and the Variation B site is converting at 12%.

Here’s a breakdown of the results…

Traffic x conversion x price = $

500 x 30% x $3 = $450

500 x 12% x $9 = $540

If you were in Bob’s shoes what variation would you keep? What variation would you tweak?

Let’s look at another example where you have a conversion button. This button has a color that fits perfectly with your branding scheme and is professionally designed to fit seamlessly in your site. Let’s split test that with a button whose color is the polar opposite of your theme, a real eye sore that you can’t help but notice. Which one does your gut say will convert better? Depending on your audience, industry, and conversion metric, results will vary wildly.

This is a great example of where all conversions are not created equal. The first example in root beer sales is pretty straight forward to measure: how many sales happened?

This conversion button is a little different and takes a little more care to digest. Be wary of an agency telling you “Look! You have 3000% more conversion on this bold button I created for you! I know it’s an eye sore but you can’t argue the numbers, I mean 3000%! Wow, just wow…” Why be skeptical? Well, unless you had a middle school intern design and create your website (they are out there, and some do decent work, relatively speaking), having a huge increase in conversions like that should raise an eyebrow.

Either this agency doesn’t know what they are doing, or worse they know exactly what they are doing and are trying to use the smoke and mirrors to deceive you in their reporting. Yes, there is undeniably an increase in conversions, you can see that in your analytics (You are running analytics, right?), but where are those conversions trickle down to you key metric of leads, sales, or other goal? Maybe your “webmaster” made the conversion button more bot friendly as well, meaning that spam bots that crawl the internet are no longer being blocked and are now counting towards your conversion numbers.

The bottom line is to keep focused on the big picture when doing these tests, an “improvement” only matters if it impacts your business.

Anything and everything on your website can and should be split tested from pricing to padding. If you aren’t actively running two version of your website and consistently iterating for improvements, you’re doing it wrong!

If you’d like to hear more about advanced tests and actual results let us know by showing us some love in the comments below!

Ken Maesaka


Leave a reply

Your email address will not be published. Required fields are marked *