Conversion Rate Optimization

The essential steps to follow for a successful A/B Test

A/B Testing is the easiest and safest way to make changes on a website, without having any programming skills. Testing changes on real customers, in real time, by splitting the traffic will show you which is the best way to act further. Though, when deciding to test hypotheses, data should be carefully analyzed, especially considering the costs of change.

In order to do A/B Testing properly, you need to know what steps to follow. The most important steps are: analyze data, formulate hypothesis and test hypothesis, measure results and then choose the solution that has a positive impact on conversions.

1. Analyze all data you have!

A/B Testing big data


If you’re sick, you need to know what’s wrong, you need a diagnose so you can get the right medicine. It’s the same situation with A/B Testing. First of all, you must analyze all the data you have to identify what are the problems on your website. Knowing and understanding the real facts that have a bad impact on conversions will help you come up with the right solutions.

There are 3 types of source that can provide relevant data for your analysis:

  • Your instinct and intuition. Sometimes, that first thing that comes to your mind is the best way to act or react. You can include in this category all the know-how you have accumulated through your lifetime experiences.
  • Web Analytics.Using web analytics for your website is the first step in having a valuable amount of statistics. Setting up a web analytics software that reports data correctly is the first step in CRO.
  • Customer Voice. The best way to figure out what visitors like or dislike on your website is to directly ask them. You can find out why your visitors leave your site without buying using surveys or feedback forms. Start asking and you will get answers! Instant chat can really help you obtain rapid information if you’re under time pressure. 

Once you have looked into data, you can easily see what the problem is. Depending on your results, you might have to change variations on your landing page, product page, cart page. You can test headlines, images, fonts, colours, call-to-action buttons and everything else you think you have to.

2. Formulate hypotheses and test!


A hypothesis is based on the insights about your customers, and it helps you see if your change can bring more conversions. When formulating a hypothesis, a marketer should include information about time, statistic relevance and cohorts. For example: “If I change the colour of the Call-To-Action button for 14 days, on 50% of the x cohort, then conversion rate will increase by 0,8 %”.

Make a list of hypotheses and then choose the one that goes perfectly with your situation. For a better selection, you need to measure the cost of your change. The cost of change is a crucial factor that can determine whether you implement or not an idea. It has 3 major aspects to consider: analyzing cost, testing cost and modification cost. If costs are too big, the change is not profitable.

A good idea is to make a top of testing hypotheses and select the best one according to 3 factors:

  • Potential. Measure the growing potential of your hypothesis and see what’s the maximum profit it can bring.
  • Effort. Value the effort it takes to change something on your website.
  • Time. If it takes too long to make a change, you should see if it’s worth doing it. Remember that time is money!

The 3 factors are interdependent, and they work together. It’s essential to analyze all 3, not only one of them. Sometimes, you have to prioritize them. For example, if an idea takes some time to be implemented, but it has a great potential to grow your conversion, you should consider it.

Of course, an idea can take no effort and very short time to change, but it has almost 0% potential. In this case, it’s better to find another hypothesis that will have a greater positive impact on your conversion rate.

After choosing your hypothesis, the next step is to test it. CRO experts recommend to test an idea for at least two weeks. This is the shortest period to have realistic and reliable results for your A/B Test. If you will only test it for a few days, you will not get useful information and you will not see the difference between the control and variation version.

3. Measure your results!

You have analyzed data, found the problem and the solution, formulated and tested hypothesis and now it’s time to measure the results you’ve got. Look up the numbers on your variation to decide if your change has increased the conversion rate.

If your numbers are bigger, if you have more newsletter subscriptions, downloads or sales, then you know you have made the right change. In conclusion, A/B Testing shows you how to bring more conversions, therefore, more sales and bigger profits. Furthermore, if the variation is more appealing to your customers, it will easily transform your visitors into shoppers and, finally, into buyers.


Results are not always very fulfilling. If you have a poor increase in results you might need to adjust your variation. There could be an error with different elements involved in the experiment. Don’t panic and find a better way to test the elements you want. Practice makes perfect! Or, in this case, testing makes perfect!

4. Implement the best solution!

The final step is to permanently implement the change on your website. For the time being, you’ll make the best of your change. As time goes by, new trends will come, consumers’ behaviour will change and will have to adapt. Using again A/B Testing, will find new, realistic and innovative solutions that fit perfectly to your customers’ needs. Never stop testing ideas! It will certainly help you improve your website step by step, based on accurate conclusions about your visitors and customers.

A/B Testing design

Leave a Reply

Your email address will not be published. Required fields are marked *