What is Split (A/B) Testing?
Split Testing or A/B Testing or Multivariate testing (doesn’t really matter which term you use, they all mean the same thing) is one way for a franchise to figure out how to best design their website or test any strategy for that matter.
The basic concept is that you can conduct experiments where a certain % of your traffic (usually 50%) is shown one version of a website while the other 50% of your traffic is shown a different version of the same webpage.
The cool part about split testing is that it is almost certainly not detectable by the end user. Since they are only seeing a single version of the page, they’re just going to behave “normally.”
So you, as the franchise in charge of the website, can sit there and watch as person after person interacts with your website, allowing you to collect data on which page performs better.
Here’s a video we like that helps explain it…
But what does it mean to perform better?
In essence, you are going to try and optimize for some sort of event i.e. How do I get the highest percentage of my visitors to do something? In franchise lead generation, the most common event is simply filling out and submitting a form.
So the question becomes, “What should my webpage look like such that the most visitors fill out my contact form?”
For example, should the page look like this?
Obviously this is a pretty drastic example, but what are some things that you can test?
- Visual elements: colors, images, videos, etc.
- Wording: headlines, CTAs, other copy
- Arrangement: Positioning of things, size of buttons, menu, location of the form
As well as little things like this, you could test more dramatic things. For example, if you simply include a phone number for someone to call instead of a contact form, how many people call in?
After you design and implement an A/B test (with the help of a computer programmer or software service), the rest is easy…just sit there and wait! (Well, you should probably work on other things, but you get the idea.)
After sufficient time has passed (usually enough to achieve statistical significance), you can evaluate the data.
Here’s a sample of what the results of a multivariate test may look like:
|Number of Visitors||37,523||37,335|
|Contact Forms Completed||378||326|
So in this instance, the "Get Free Info" button performed significantly better than the "Learn More" button.
For an example of one of the A/B tests we ran at FranchiseHelp, take a look here.
If you have any questions about A/B testing shoot us a note here.