• info@helpingtesters.com
  • helpingtesters

What is A/B Testing and How to Perform

March 10, 2017 Types of Testing
a/b testing,a b testing, ab testing, ab testing tools, a b testing tools, ab split testing

The software testing and mobile application testing concepts ensure the effective performance of the application before release. When considering the online promotional needs and marketing strategies the A/B testing is an excellent method for testing the performance of two versions. A/B testing tests the two web pages by comparing the variants of the same web page released to similar visitors at the same time.

Checks for the better conversion rate and select the one which performs well. A/B Test plan is extremely different from other Software Test Plans. The A/B test plans defined well will be effective to identify the insights about the customers and the conversion rate. Also, this will be useful for detecting the exact preferences of the customers.

What is Conversion Rate?

For web sites like an e-commerce website, a website for a service industry, travel agency website and much more … the purpose is to get goal achievers. The quality and style of the presentation affect the mindset of the customers. The visitors should be changed into the users of the services offered. The rate of converting visitors to the goal achievers is defined as the conversion rate.

AB Testing is much related to the way in which the product/website is to be marketed. The testing concepts are not much popular and people are less aware of the wide possibilities of A/B Testing. So, A/B testing has not become as common as SEO, Web Analytics, link building etc.. The fact is that A/B Testing will be effective to increase conversion rate and evaluate the user behavior. Two or more versions can be designed for the evaluation and two versions are hosted first for testing. The most effective version will accept and continue the testing with other versions designed. The version selected at the end ensures a successful behavior in the real-time use.

How to perform A/B Testing?

For performing A/B Testing on the web, two versions of the website is designed, the existing version called the Control and the new version called the variant. Two or more variants can be shown in random to the users and test for the impact on user behavior.

For a website or mobile application what to test will be determined based on the business goals.
If the website is an e-commerce website you may need to present the products by explaining the details in brief. People may not interested in struggling to find out the details of a product. The presentation of the products should be attractive and should make them visit the details page.
This can be done by:

  • More attractive and clickable buttons.
  • Placement, Colour, Size and wording all are important.
  • Landing pages with perfect images and descriptions.
  • Catchy Headline and Product Description.
  • Responsive style and layout of the website.

The presentation styles can check by randomly distributed versions for a similar group of customers. The primary process of A/B testing will include the above points. Along with these points, the testing process will change depending on the goals and based on the insights generated there might make effective changes in the versions.

Steps to design an A/B Testing

 

Data Collection

The data collection is the process of identifying the high traffic areas of the website or the mobile application. The data optimization can start from the pages where conversion rates low or drop-off rates high. This analysis can be done from the analytics reports.

Designing the Goals

Based on the requirement of the website/application the goal is designed for creating test framework. The increased traffic in the website can be achieved by well-defined functionalities of the website, developed based on the goal identified. The goal can be anything like a link to purchase page, clicking a button, registration/sign-up form, social media networks links etc…

Generating the test Ideas

Based on the goals identified the next step is to design the A/B Testing ideas. Design the test plans for each function and prioritize them based on the expected impact on user behavior.

Design and develop the variants

The variants are designed based on the test ideas developed. The changes can be made by alternating the colors of a link or button, shifting the style of a control, reciprocating or reordering the features listed on the website. The impact of the customer behavior will vary on small changes in the appearance.

Execute
Host the two versions (A and B) and wait for the reports. Should assign randomly the control and variation to the customers and test their behavior upon both versions.

Result
The test results are analyzed to determine the effects of the variants and control. The results analysis can be done by using tools like Optimizely and VWO. The tools come with built-in Google Analytics integrations.

A/B Tests with Google Analytics

 

A/B Test with Google Analytics is the easiest way to design A/B Testing. The best available tools for A/B Testing are paid and so Google Analytics integrates a perfect way to analyze the results of testing. The testing will advise the users to determine whether a single conversion goal is enough or not. The analysis will help to develop multiple conversion goals based on the requirements. The screens below explains how to process A/B testing.

Login to Google Analytics
Choose the Behaviour – Experiment

a/b testing,a b testing, ab testing, ab testing tools, a b testing tools, ab split testing

Created a new Experiment for Comparing Landing Page Style and Design.

a/b testing,a b testing, ab testing, ab testing tools, a b testing tools, ab split testing

‘The objective of the experiment’ can set as ‘Bounces’. There are many other options to select as the objective. The metrics can be such as Ad Sense, e-commerce, Goal metric, Site Usage etc… Also, there is an option for creating new objective as per the requirements.

‘The percentage of traffic experiment’ can be selected to 50%, which helps to make 50% of the visitors should see the original and another 50% should see the variant.

  • AdSense – To improve ad clicks or impressions.
  • E-commerce – To boost the revenue.
  • Goal metric – To manage session duration, destination page clicks.
  • Site usage – To evaluate average page views and time on site.

 

The variant is also prepared and given for comparison.

a/b testing,a b testing, ab testing, ab testing tools, a b testing tools, ab split testing

The next step will generate the Experiment Code. Copy the Experiment code and paste immediately after the opening of the head tag in the original web page. When the page is updated click on “Next”.

a/b testing,a b testing, ab testing, ab testing tools, a b testing tools, ab split testing

Click on the “Start Experiment” and Google will provide a report within next 24-48 hours. This will help to make an analysis of the performance of both pages [Version A and B]. Also, it is possible to add more variants [variant1, variant2 etc..] for making more analysis.

Conclusion

To conclude, A/B Test result analysis helps to determine the minor issues of the websites which may lead to reduced page rank. The landing page efficiency checks will help to improve the efficiency of the website. The Statistical significance is a technical term which measures the rate of conversions to the visitors. This factor is calculated in percentage. The data collected from the report will be useful for the preparation of further variants for repeat testing. Unfortunately, due to the reasons like some software testing myths, unawareness of the A/B testing etc.., most of the software companies keep away from A/B testing. Nowadays Google analytics methods, SEO tactics are used widely for marketing. A/B testing will also contribute effective marketing tactics by providing exact conversion rates.

About the author

Meera Sandeep author

Leave a Reply

Your email address will not be published.