What’s an AB Test (or Split Test)
An AB Test, also know as Split Test, is an experiment which consists in comparing 2 or more versions, in order to understand which one works the best.
On what can an AB Test be done? A Split Test is usually done between 2 web pages, however this method can be also applied, as you’ll see, to newsletter, advertising, pop-ups for email addresses research.
Let’s see in details on what an AB Tests can be done and with which kind of goals.
AB Test or multivariate Test?
Unlike the AB test, a multivariate Test can be used to test different elements on a page.
A multivariate Test isn’t about testing 2 different versions of the same page, but about testing different versions of the elements inside a single page.
A multivariate Test is useful to understand which elements of a page play the major role on making you reach your goals.
7 elements that you can test using an AB Test
An AB Test isn’t only about a website or a landing page: everything can be tested, from emails to advertising campaigns.
In this section, we’re about to discover 6 elements that you can test using an AB Test, that is how you can:
- AB Test 2 or more web pages or Landing Pages
- AB Test 2 or more subjects (and senders) of a newsletter using Mail Chimp
- AB Test 2 or more pop-up of email collection using Optin Monster
- AB Test using 2 or more Facebook Ads
- AB Test using Google Adwords
- AB Test using WordPress
- AB Test 2 Google Adsense ads
AB test of 2 or more Web Pages or Landing Pages
As said before, the most classic application of an AB Test is between two web pages or two Landing Pages on a website.
In this case, the same visitors’ number is brought to two different version of a web page or two completely different pages (the A one and the B one indeed) in order to see which one, the conditions (or visitors) being equal, better works under many points of view.
There’s a typical example: as you can see
- 50% of visitors is readdressed to the A Version of the site;
- The remaining 50% is readdressed to the B Version;
- A Version is the one that works better.
Do you get it? With conditions being equal, an AB Test or Split Test, lets you verify which of the 2 version of a website page works the best.
Under what terms? In other words, what are we trying to verify when we compare 2 versions of the same webpage?
When an AB Test is made there are some general goals and some other specific ones.
In this case, the general goal might be the conversion, our AB Test goal is to verify which version converts the most, whether this means selling more (like in the case of an ecommerce) or being able to gain more contact records (as in a lead generation).
AB Test of 2 or more subjects (and senders) in a newsletter using Mail Chimp
What’s the element that most effects the addressee in opening your email and read it or trash it? The subject of the email, beyond the sender, which are the only 2 elements that the addressee sees when he’s receiving an email.
Ok, but what’s the subject the works best, that makes someone open the newsletter you’re sending? There are many techniques of improving the effectiveness of your email subject in order to make them to be read, but you know:
What is the best method to find a subject that makes someone open an email? AB Testing it.
How to do a Split Test using Mail Chimps in 4 steps
Yes, you can AB Test using any email marketing software: let’s see how to do it with Mail Chimp in 4 steps.
First of all you can decide if testing two or more:
- Email subjects;
- Sending time;
Once you’ve decided what do you want to test, you can decide how big the first group you’re sending the A and the B version is: Mail Chimp, by default, suggests you to send using a 50%, which means that you’re going to send the A version to the 25% of your contacts and the B version to the remaining 25%.
Mail Chimp will determine the winning version after a few hours, 4 by default, but you can choose how long you’re going to wait: Zapier suggests you to wait 4 or 5 days before determining the winner of your AB Test.
Once determining the winner version, the subject that generated the higher number of openings, for example, Mail Chimp will send it to the remaining group of visitors (for example, 50%).
AB Testing 2 or more collection emails pop-ups using Optin Monster
For example, you can test:
• Button’s colors;
• Pop-ups texts;
• Number of fields;
AB Test of 2 or more Facebook Ads.
You can obviously test 2 or more ads, for example Facebook Ads or Google Adwords for example, in order to see which one works the best in CTR terms.
Thanks to an AB Test with Facebook Ads, AdEspresso made its conversion rate and CTR rise, and its cost per lead decrease:
You can execute an AB test using Facebook Ads thanks to tools as:
- Adespresso itself.
AB Test using Google Adwords.
You can split test your Google Adwords ads
- On a search network, alternating similarly two ads, comparing them with a tool as Splittester by Perry Marshall.
- On a display network using Perfect Banner;
AB Test using WordPress.
You can AB test also using WordPress and its plug-ins such as:
AB Testing 2 Google Adsense ads
An AB test lets you testing two Google Adsense ads in terms of performance: going to the “my ads” section and clicking on “new experiment”, you can create a test that compares the original ad with another version of the same ad.
Google Adsense will alternate both in order to understand which one, the views being equal, is the most clicked, so the most profitable.
5 fundamental elements that you must test in a Landing Page
How come, in the last example, A version works better than the B one?
In other words, which elements you must test and which changes you must make to the page so that it converts more?
Here you are a list of the typical elements that you can test in terms of performance:
The headline is the fundamental part of a landing page. In order to make you understand the importance of testing the headline, statistics confirms that:
- A strong headline can rise the effectiveness of your content up to 500%;
- The headline can increase the conversion rate of a website up to 40%;
The Highrise case: +30% conversions
A well known case of application of an AB Test on headlines is the Highrise software case, that tested two different headlines versions, discovering that “30 days of free trial for all the accounts” generated a 30% subscribers growth compared to the old “Open an Highrise account”
The CareLogger case: +31% conversions
Another example of the applications of an AB Test on a headline, comes from CareLogger, which has increased the conversion rate by the 31% simply changing its headline:
What’s the reason behind this? Probably, as we’re about to see with another example, shifting the focus from the disease (“Diabetes”) to the “Optimal Health”.
The subtitle, on a landing page, takes the problem that the headline has already shown and gives the solution to it.
An AB test on the subtitle of our landing page lets us understand which one has the best performance. The demonstration comes again from Highrise: Same headline, different subtitle, allowed an subscribers tax increase from 27% to 30%.
It was enough changing the subtitle from “Pay as you go. No Long term contracts. No Hidden fees. No surprises” to “Signup takes less than 60 seconds. Pick a plan to get started”. How can we explain this improvement?
That probably happened because of the subtitle changing from a negative expression, to the talking about benefits, with a clear call to action.
Grabbing the user’s attention on potential problems (long-term contracts, hidden fees, unpleasant surprise) in the subtitle, even just to deny them, might have had a worst effect than just moving the attention on a clear benefit (you can signup in less than 60 seconds) and on an as much clear call to action (Pick a plan to get started).
Call To Action
A Call to Action is that sentence in a landing page that gets you to do something, buying the product or leaving your contact information, as it happens in a Lead Generation strategy, in a landing page’s particular context known as Squeeze Page.
Soocial: from 14.5 to 18.6% conversions.
Can a simple word adding increase the conversion rate by the 28% on a call to action? It seems like it can, at least according to the AB Test that Soocial made: the simple adding of “it’s free” increased the conversions percentage from 14.5% to 18.6%
Testing your call to action is not necessary to obtain benefits: it’s enough just to change the color of the button, as “CareLogger, which has improved by 34% the conversion rate, simply testing red and green buttons” showed
A similar result comes from Performable, which testing two versions of the call to action button, has had a 21% CTR increase on the red button:
What does the last AB Test teaches us? Red is always a good choice?
No! It teaches us not having prejudices: red is traditionally associated to danger and mistakes. Only not having prejudices and thanks to a split test on these two versions, Performable did successfully find the most performant version.
Length of a Landing Page
Does it convert more a short landing page or a long one? It depends. And the best way to found it out is an AB Test.
Despite this suggestion, a split test, allows us to understand the right length that we should use for our web page.
Moz succeeded in increasing the number of the subscribers to his software up to more than 1 million dollars per year, working on some modifications that involved the length of the selling page too:
Another element that can be tested using an AB test on you webpage is the contact form.
What can you test in a contact form? Here’s some ideas:
Test the quantity of fields that you request
Statistics confirm that the more information you ask for, the less people give.
You could test:
- 2 different versions of your contact form, one with more fields and the other one with less;
- A split contact form, in other words, divided in two: following the rule Commitment and Consistency, when people start a job, they tend to finish it.
Neil Patel has succeeded in increasing his contact form’s tax of conversion of 27%, by testing two different version of it, one containing 3 fields and the other one containing 4 fields, simply reducing the “online revenue” one:
Test your contact form position
Where do we put our contact form on a landing or squeeze page?
Someone would say above the fold as a obvious answer
An AB Test made by Content Verve has brought an increase of conversion by the 304% just by positioning the contact form on the bottom side of the page.
How to AB Test using Landing Page
How can you make an AB Test or Split Test using a Landing Page?
Basically there are 2 ways:
- Specialized softwares as VWO, Google Analytics or Optimizely;
- AB Test tools provided by some websites that help you in creating landing pages;
Let’s see the first option
3 Softwares for AB Testing
Which software can help us in our AB Testing?
Starting from Google Analytics, VWO and Optimizely, let’s see a list of the 3 major split testing softwares.
How to create an AB Test with the experiments of Google Analytics in three easy steps
Google Analytics allows you to AB Test by the creation of experiments in few easy steps
This functionalities have replaced Google Website Optimizer in 2012
First step is going on the Analytics/Experiments section
Once clicked on “create a new experiment”, it is necessary to insert some fields such as:
- the name of the experiment;
- the goal of the experiment, or the metric according to what the experiment is valued on and the winning page is chosen;
- the possibility of uniformly deliver all of the traffic on all the variants, to assign an equal quantity of traffic to each variant, for all the duration of the experiment;
- the minimum duration of the experiment’s execution: Analytics will not chose a winner before this time expires;
- a threshold of reliability which has to be reached before Analytics can declare a winner
Through the next section it’ll be possible to test the variants and the monitoring code of the experiment:
As well as Google Analytics, you can use other softwares in order to create split test: you can find a complete list of all the tools used to create an AB Test here, anyway the most known are:
Visual Website Optimizer is an easy tool used in order to AB Testing something, which allows you to create different versions of a website or a landing page, using an editor, and see which one makes more conversions or selling.
VWO also allows you to create multivariate test, heatmaps as well as other 1000 functionalities.
You can free try VWO for 30 days.
Optimizely allows you, as VWO does, to create an AB Test or a multivariate test.
How to calculate when a Split Test is meaningful?
Let’s imagine you having tested two different versions of your landing page using a split test or a multivariate one.
When the obtained results may be considered reliable? In other words, how can I verify the validity and the statistic relevance of my test?
Two tools can come in help: let’s see them.
Two tools useful to calculate the validity and the statistic relevance of a split test
It lets you know if the obtained result is statistically reliable or not, from the number of visitors and conversions.
How long should our AB Test last to be significative?
This tool gave us the answer, simply putting some datas as the estimated tax of conversion, the number and percentage of the visitors included on our test.
Other experiments and AB Test to consult
Online there are to be found a lot of AB test, anyway, to get deep in the subject, I suggest you to read the following articles:
Now it’s your turn
Did you make any AB Test? Which results did you obtain? Let’s talk about this in the comment section!