6 Essential Steps for Successful A/B Tests

/, Google Analytics/6 Essential Steps for Successful A/B Tests

6 Essential Steps for Successful A/B Tests

In the online marketing world, A/B testing has become the staple of website optimization, which leads to higher conversion rate. However, many online marketers or small business owners are intimidated by the complexity of A/B testing, multi-variable testing or any testing on their websites. It doesn’t have to be hard.

In order to become successful in A/B testing, you have to run as many tests as possible and have as high impact per successful test possible.

It boils down two things:

the success of ab testing

The number of tests + the percentage of tests that provide a win. Don’t forget the average sample size and its impact.

Another thing to keep in mind is that do not focus on tactics, things like make the button bigger, write a better headline…etc, focus on the process instead.

Before starting your actual A/B testing, make sure to assess your website for the following criteria.


Does the page meet user expectation?


is the content or the offerings on this page as clear as possible?


Is it communicating value to the user?


What on this page is causing doubts, hesitations and uncertainties?


What is on the page that is not helping the user take action?

Technical Analysis

After you have gone through the above process, you need to do a Technical Analysis. Done correctly, this can boost your conversion and make you a lot of money.

Conduct Cross-browser and cross-device testing

To start, go to your Google Analytics account, navigate to Audience -> Technology -> Browser & OS report. You will see conversion rate for each browser. Make sure you only look at one device category at a time: desktop only, tablet only and mobile only. You can drill down to a specific browser version, for example IE8, IE9…etc, to see if a particular browser converts less than others.

Use the following tools to find the issues:

  1. Crossbrowsertesting
  2. Browserstack

ab testing tools

If you find a low-performing browser that’s not used very much, look up the number of the visitors for that browser, and determine whether it is worth the ROI to identify and fix the problem.

ROI of fixing the browser bug

Go through the above calculation, if ROI is greater than or equal to 0, then you should definitely fix the bug. After six months, you will have a positive ROI.

Speed Analysis 

if your site is too slow, it might not only hurt your user experience, it will also hurt your search ranking.

To find out your page load time and page interactive time, you should go to your Google Analytics, navigate to Behavior -> Site Speed -> Page Timings. From there, turn on the ‘comparison’ to easily spot slower pages.

If your page speed is from 0 – 3 seconds, you are good.

Related:  The Evolution of JCPenney Brand

If your page speed is from 3 –   7 seconds, you are still good, but definitely can use some improvement.

If your page speed is from 7 to 10 seconds or longer, you should spend some effort at improving the page speed.

page loading speed

To start, identify the pages with the most traffic and slow page load time and page interactive time per page.

Use Google PageSpeed Insights and enter every URL you had identified from the step above.

Web Analytics Analysis 

Web Analytics is critical when it comes to identify what people are dong on your site, and the impact and performance of every feature, widget and content on your site.

Make sure you draw insights from your web analytics data that can help improve your website, instead of just looking at it.

Qualitative Surveys 

You can also survey your web traffic to figure out if there’s any user friction they experience on your web pages. 2 ways to survey your web traffic:

1. Exit surveys: implement a popup when they are about to leave your site

2. On-page surveys: ask people to fill out a survey when they are on a specific page.

A very useful tool is Qualaroo.

survey tool

To start, you should identify the pages that you want to have your survey, then set your own questions depending on your own needs, do not use the template. Then you can set your criteria for when the survey request would be initiated.

With any user testing, make sure to observe actual people using and interacting with your website while they are commenting their thought process out loud. Don’t force them or talk to them, just observe and listen to their commenting.

Make sure to include these 3 types of tasks in your test:

1. A specific task

2. A broad task

3. Funnel completion

fix ab testing

Time to Put Everything Together and Fix!

After you have your findings, remember to categorize them in different buckets.

some issues are low hanging fruit, which means the issue is obvious, and requires minimal effort which will result in high conversion rate.

If there are obvious opportunities to change behavior, or increase conversions, this bucket is for standard tests.

If you have found a page, or widget that is not working well, but don’t see any solution, put it in the Hypothesized issues bucket. We will go back to that when we have time.

High-level issues are those that need to do some more testing with particular devices or need more information to finalize the problem.

Lastly, the instrument bucket should include problems that involve fixing, adding or improving tags or analytics collection to improve insights.

ab testing success

Good Luck with your testings!

2018-03-01T02:40:01+00:00By |Categories: Digital Marketing, Google Analytics|Tags: , |

About the Author:

Cofounder of Siteber: a digital marketing professional with Master of Science in Marketing Analytics, specialize in SEO and Digital Analytics.

Leave A Comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.