All that you need to know about A/B Testing
Why we chose this topic is simply because it is not used much as the internet marketing. This is a great tool if you want to understand the behaviour of the visitor and change the visits to conversions. What happens with A/B Testing is that people do not understand what it is and how it can benefit the user. We hope that in this article we will be able to explain to the readers more about what the A/B Testing actually is.
What exactly is A/B Testing?
Well as the name explains. A/B Testing is basically about testing two versions or alternatives. You check both the alternatives and then decide on which alternative was better. Both the versions must be experimented simultaneously only then you will be able to judge which version is actually better
It is basically like any science experiment that you must have done. For example you want to study what are the things that promote the growth of a plant and what are the factors that do not allow the plants to grow better. This can be done when you grow the plants under different conditions and try to find out what makes the plants grow better.
Now let’s look at an example from blogging point of view. Suppose you have a website that has two designs. You make some visitors see one design and the other visitors see the second design. Using metrics you can easily see which variation is actually leading to more conversions. The enclosed diagram will explain what we are trying to say here.
What are you testing?
Each scenario will require a different A/B Testing. However, some elements that are usually
tested are as follows:
- Amount of text on the page is sufficient or not.
- The lengths and types of fields that the form contains are fine.
- Layout and style of the website is fine.
- The images on landing and product pages are fine.
- The headline or the product description is adequate.
- The call to action is fine – for example the size of the button, the color and the placement as well as the wordings.
These are the common things that are tested with the help of A/B Testing. Apart from these you can customise the A/B Testing and check other parameters too.
How to create the first A/B Test
One thing that you must know here is no matter which tool you are using the basic set up remains the same. Either of the below mentioned two methods can be used:
Redirect to another page – Suppose you want to do A/B Testing on a whole page. For example we want to test a blue theme and a pink theme. All that you will have to do is create and upload a new page on your site. As you are experimenting between two variations you will have to create a variation location. As the test is running the tool will automatically redirect some of the visitors to the other location so that you can see the reaction of these visitors.
- Replace the elements that need to test before the page loads – Another method can be used when you are testing a single element on a web page. For example you are testing the sign up button. You will have to create only variation of this button using HTML as your testing tool. As the A/B Testing goes live the tool will automatically replace the original button that was there on the page with the new variations that you have created before the page is displayed to the visitors.
After a number of such conversions happened the website owner can check which variation actually gave the best result. According to the results you can make the necessary changes on the site.
A/B Testing is definitely easy and can be used by anyone.
What you must keep in mind during A/B Testing
A/B Testing is based on a very simple concept and anyone can successfully use it. However, there are some points that you must keep in mind as you are doing the A/B Testing. This will ensure that the possibilities of you going wrong while doing A/B Testing are reduced to the minimal.
You must know for how long you must run the test before you give up. Some people lose patience and they tend to leave very early. This can be a problem and can lead to wrong results. Similarly if you keep testing for too long the results can be equally misleading and can cause you loss of sales due to variations that are not preferred by the visitors. There are some tools that are available to understand that for exactly how long you should do these testing.
The tool must remember which version of the variation was shown to the visitor. This will ensure that the visitor is shown the same version again when he or she is visiting else it may lead to a lot of confusion.
The A/B Testing should be consistent on all the pages of the site. For example if you are showing one variation of the “buy now” button to the visitor you must see to it that no matter to which page the visitor goes he or she gets to see the same “buy now” button. If the visitor gets to see different versions on the same button in different page he or she will get confused and it will affect the results of the experiment.
- The A/B Testing can have three different types of results like a positive result, a negative result and also no result at all. The best way to get the positive A/B Testing results is to keep doing the tests again and again. This is the biggest fact about A/B Testing which you must remember.
Things that should be avoided during A/B Testing
We will now talk about the things that you should never do during an A/B Testing process. This wills help you to ensure that the whole process of A/B Testing goes on smoothly.
- As we have mentioned before too the
A/B Testing should not be concluded too early. There is a concept of statistical confidence this will tell you if you should take the results seriously or not. Only a few visitors studied cannot give you the true picture. A/B Testing tools will come with the feature of statistical confidence. If you are doing the study manually then you might have to check it yourself.
- Both the versions that you are testing must be tested simultaneously. If you test the two versions are different time the results will never be correct. The traffic between the two versions should be split equally else the results will be misleading again.
- Never go by what your gut feeling says as most of the time the A/B Testing will give a result that you were not expecting at all. You never know a green button on a red page may actually give you a much better result than you thought. You must look for the variation that works best with your visitors and not for a variation that you like. Basically do not be judgmental have an open mind.
- Another mistake that people often do is that they guide the regular visitors to the variations being experimented. A regular visitor might be confused to see a new variation especially if the variation is later on not implemented. Try to use only the new visitors to check the results of the experiments.
A/B Testing case studies
There are many case studies that can be associated with A/B Testing. We cannot explain all of them in one article. We will try to explain one such case study here.
It was observed that human images can increase the conversion rate as visitors are more attracted to images with people.
Google Website optimizer case study was conducted by Tim Ferris and it was realized that when visitors were given lesser options to choose from it resulted in more conversion. In this case the winning version was also very easy on the eye of the visitor.
Another interesting study we must mention here is the use of the word “It’s free”. This had increased the sign up percentage by 28%. Another example here would be 37Signals they changed the headline from “start a high rise account” to “30-day free trial on all accounts” and this increased the conversion by thirty percent.
CareLogger increased the conversion rate by simply changing the color of the sign up button from green to red.
Some tools that are often used in A/B Testing
- A/Bingo and Vanity – This is a server-side framework that is used for the Ruby on Rails developers. This requires a lot of programming and integration in code as a result might not be suitable for all.
- Google Website Optimizer – This is a free
A/B Testing tool from Google. It is a good idea if you are trying to learn about A/B Testing. In case you are already into A/B Testing and looking for advanced features then this is not what you must use.
- Vertster, Webtrends Optimize, SiteSpect and Omnitures’s Test&Target – These are nothing but enterprise testing tools.
These are some of the points on A/B Testing that we thought must be mentioned. This article is basically about giving an idea of A/B Testing is all about. There are many online materials and articles that you can read if you want to understand more about A/B Testing.