A/B testing Process explained from the scratch

In the past few years, there has been a paradigm shift from conventional marketing to digital marketing. This is mainly because people all over the world have started consuming online information more than anything and digital marketing turns out to be one of the best ways to reach out to target customers. Email marketing, social media marketing, SEO, etc are all different ways of digital marketing methods which are capable to create an effective communication channel with consumers and offers high conversions. The conversion rate is the percentage of viewers who get converted into leads or subscribers and in the end finally purchasing your product or service. Higher conversion rates imply your marketing methods have been successful.
So, every marketer is trying to figure out the best way to attract customers and increase conversion. But how do you know which of your strategies – be it a landing page or a sales email or search ads works best in attracting customers? This can be usually understood by monitoring your customer behavior- how the visitors of your page respond. You might have to consider changing strategies based on visitor response than for optimization. But how do you know if your changed strategy is better than original? This is where A/B testing comes in. A/B testing is a method for figuring out the best marketing strategies for your business. In this article, we will be seeing what is A/B testing and how it helps you with your ad campaigns or promotional pages.

What is A/B testing?

A/B testing is the new technical word for your old school controlled experiments. Most of us would have done a controlled experiment in your school days in Science lab like growing bean seeds in two different pots, with one varying condition- like not exposing one pot to sunlight or not watering a pot, etc. These experiments helped us understand the factors contributing to seed germination. We could come to decisions like a seed needs water and sunlight to germinate and grow or so on. A/B testing is a simple technique in which two versions of a web page are compared to determine which one performs better. The versions would have different design elements and the results of A/B testing would give an idea what all changes on the web page elements helped gain customer attraction and increase conversion. It can be done to any web page- a copy of your website, your e-commerce site, your landing pages, etc.
So, to put it simply, A/B testing involves testing an original design, A (usually called the control) against an alternate version, B (usually called Variant) to see which performs better. More than one variant may also be included in the testing process. The pages are shown to users randomly and the performance of different variants is analyzed using statistical analysis to determine which one performs better for given conversion goals. To carry out A/B testing, the traffic is split between the different variants so that each variation gets a random sampling of visitors. For this reason, A/B testing is also known as split testing.
A/B tests could be carried out for off-site as well where you could probably test an ad or a sales email. Testing two versions of ad copies or emails to see which one brings in more converting visitors can help you determine which one works better.

What to test in your A/B tests

You could test any of your web page elements like the headline, image, video, etc in A/B testing. Usually, it is easier if you pick one variable to test in a variant and measure its performance. This makes it much easier to evaluate how effective a change is. Otherwise, you might not be able to identify which change or element was responsible for changes in performance.
But there are also instances when it makes more sense to test multiple variables rather than a single variable. This is called multivariate testing. Multivariate testing is mostly used when you want to know how the different elements interact with one another. But if you are trying to narrow down the most effective elements, the best possible solution is to do A/B testing by testing one element at a time on a variant page. Even small changes in your elements, like changing the words of your Call to Action (CTA), or adding an image or video can drive in huge changes to your conversion.
Now, coming to what all elements can be A/B tested, the answer is anything that you feel could affect customer behavior. Few of the elements that you can A/B test on your webpage are:
• Titles and Headlines
• Sub-headlines
• Copy text
• Call to Action text
• Call to Action Button
• Images
• Videos
• Page Structure
• Forms
• Navigation
This is just a list of things; like I already said it could be anything that you feel could affect your customer behavior. In selecting what to A/B test you have to be more importantly process minded and have a structured approach in identifying the factors. For that, you have to research, prioritize, experiment, analyze, learn and repeat. Let’s now discuss how we can achieve this in some distinct steps so that it would be much easier to digest.

Step by Step Process to A/B testing

1. Research: Gather data-driven insights on your website data and user behavior

A/B testing, as you know is an important step to optimize your webpage. So, to start with the optimization process you first need to identify what your customers want and how they respond to your web pages. So start researching. You get a lot of tools which comes handy in doing research and gathering data. Use a tool like Google Analytics for Web Analytics where you gather your website data. There are a lot of tools available in the market to analyze your visitors' behavior from Heatmaps, visitor recordings, form analysis, and on-page surveys. Even you can gather data by user testing your web page. These data you gather are very important to gain insights on your visitor behavior and identify what elements of your page are responsible for conversions or what elements of the page is stopping the visitors from converting.

2. Prioritize your goals and build a Hypothesis

You will mostly end up with too many insights from your research results. Now it’s time to prioritize your goals. You probably might not be able to test all your insights. So prioritize them based on various factors like severity or importance, ease of implementation, an opportunity of implementing the change (how big an impact can the change bring in conversions).
Now, based on your prioritized goals, build a hypothesis. A hypothesis is like a prediction you feel could bring in conversions to your webpage. You arrive at these predictions based on your research data. For example, your research data show, many users did not click your CTA button because it was not prominent on your page. So your hypothesis may be something like: Changing the color of the CTA button with a contrast color will make it more prominent and can increase conversions. Your hypothesis should clearly state what is to be changed, what will be the outcome of the change and why you think so.

3. Create a variation as per your hypothesis and test it.

Once you have created your hypotheses, create variations based on them. Now need to actually A/B test the pages. You have your control (original page) and one or more variants. There are many A/B testing tools available in the market like Optimizely, HubSpot, Unbounce, Visual Website Optimizer (VWO) etc. Split your traffic to sample groups randomly and equally among the different pages. Many A/B test tools do this automatically so that each variation gets a random sampling. You could also determine your sample size and split traffic.

4. Analyze your test results and make decisions based on your results.

Analyze the A/B test results to determine which variant performed well and delivered higher conversions. But before you analyze, make sure you have run you’re a/B tests long enough to obtain a substantial sample size and you have got statistically significant results. In other words, your results should be significant enough to justify the change. The duration to get statistically significant results depends on how much traffic you get to your web page. Less the traffic to your webpage you would have to wait longer. So, depending on various factors like your company, sample size, how you carry out you’re A/B tests, etc you might get statistically significant results in hours, or days, or weeks, or even months.
Based on the data from your analysis you might or might not have a clear winner among the variations. If a variant gives you more conversions compared to others, it turns out to be your winner and you can go ahead with its implementation. In case you don’t have a clear winner and your test remains inconclusive, you might have to repeat your A/B test for longer duration, or even modify or rework your hypotheses and come up with new designs for your variants.

Conclusion

That was a quick glance on A/B testing. Practically, to optimize your web pages you should keep on collecting data and A/B testing to increase your conversions until you reach your target. A/B testing definitely is a fantastic method which you should not miss in your marketing strategies in order to order to figure out the best web pages or landing pages and to improve your ROI on your marketing campaigns. Hope this article helped you get an overall idea on what is A/B testing all about started with it!