Ecommerce A/B Testing and Conversion Optimization

Getting traffic is the first part of ecommerce growth. Converting them in to customer is the second part. In order to get more revenue from your existing visitors, you need to discover how you sell your products. Guessing and “I think this is the best way” are terrible ways to find best website design or selling experience. There is a/b testing process to find out for this process and in this post you will learn everything about ecommerce a/b testing in order to optimize your conversion rate.

A/B Testing Process

Probably you have heard lot’s of different examples of A/B testing. Changing button color are most popular A/B testing types. In this post we are not going to talk about A/B tests like that. Actually button color is not that important if you are small or mid size ecommerce website.

We look at a/b testing process is convincing people to make purchase decision. This is completely different issue and completely based on sales-marketing mix. Changing forms, adding UX messages or button colors etc. etc. These can not help you to convince visitors to make purchase decision.

Did you ever cancel your order that you want to have because of a button color or form element on page? No. No one do that so skip that bullshits and learn how to convince people to make buying decision.

In conversion optimization process, most important thing is finding right scenariou to test. I’ll share a four step process we use at Growth Hacking Studio to succeed at conversion rate optimization.

1. Understanding Segmentation of Your Users

Segmentation is the heart of conversion rate optimization and a/b testing process in ecommerce website. There are lots of ways to segment your visitors. Since our topic is conversion rate optimization, so we need to segment our users based on their potential to purchase decision.

Surfers are generally come from display ads campaigns. They have highest bounce rate. Discounts, product description or amazing UX won’t affect this audience behavior. You need to ignore this segment of users.

Explorers are the gold mine for you. They have potential to make purchase but they tend to postpone. This is what makes them gold mine. If you find ways to make their purchase decision faster, then you win! You got a new revenue stream!

Heros are made their mind. They generally saw your websites at least 4-5 times before become a hero. Make compression, look at reviews and their last session is all about finding the right product and completing purchase.

As you guess, conversion rate optimization process should focus on explorers and heros. Since we know segment that we use, let’s go to analysis and research stage.

2. Analysis & Research

As you know, segmentation was the heart of conversion rate optimization process. Well, analysis and research is the brain of this process. In this section, we will learn how we find out conversion rate optimization hypothesis for these audiences.

Hypothesis? What is that?

Did you know that “I think …” is the killer of many online business growth today. What you think belongs to you. Not your visitors. Period.

So, how can we proceed this process? It is very simple.

There are NO GROWTH IDEAS. We only have HYPOTHESIS based on analysis and research. All hypothesis should be tested to be validated. In this section you’ll learn how to create conversion rate optimization hypothesis.

Since our hypothesis based on analysis and research the first thing you need to get done is building data collection methods. The first and most important data collection method is using Google Analytics.

How to Use Google Analytics Efficiently?

You start your ecommerce website via Woocommerce, Shopify or any other platforms. Find plugins to enable Google Analytics tracking. You add your tracking ID and start to see your sales and revenues in reports.

You might wonder, so what?

Yes. So what?

If you have an intent to growth your online business, please do not use Google Analytics is a report tool. Google Analytics make possible to data analysis if you use it in that way.

What is the difference?

In order to get deep insights, you need to define micro conversions like product photo/video engagement, seeing particular part of content, installment options, filtering on listing page, etc. You got the idea.

When you start to track micro conversions as well as macro conversions (transaction, sign up etc.) then you are ready to use Google Analytics for data analysis.

What You Should Expect From Google Analytics Reports?

Google Analytics reports will show you where the problem is. Pages and micro conversions you need to research. 99% of time you can not find solution of problem using Google Analytics.

Device reports, screen resolution reports, source/medium reports, landing page reports, all pages reports, user flow reports, checkout and shopping funnels are the essentials when you start analyse efforts.

You need to find an abnormal activity based on user segments that you create on Google Analytics or time decay you choose.

Let’s look at a few examples;

Landing pages that have highest bounce rates
Landing page differences based on devices
Landing page differences based on users location
Page value report based on users type (new vs returning)
Checkout and Shopping Funnels based on device, user type and user’s count of session
Etc. etc.

When you look at these reports in this perspective, you’ll find some of your pages shows different performance rest of your website. This differences worth to investigate more.

How? Research process starts here. Let’s look at most common research ways for conversion rate optimization and A/B testing in ecommerce.

Heatmaps & User Recordings: Understanding users full story needs more than statistics. Heatmaps and user recording is the most easiest way. There are lots of different tools like Hotjar, Full Story, Mouseflow, Yandex Metrica etc.

Use one of them to get more information about specific pages, events or user flow. This makes you get more informed about your analysis result in previous step.

Is this enough? No, it is not. Most of time, even this will not will be enough. If you feel, you need to get more information about this kind of issues then start surveying for your user. Surveys are great way to collect more information from your users.

Hotjar makes it easy to survey on your webpages and if you need to use your existing clients, then you can use Google Forms, Typeform, SurveyMonkey or similar products.

When start surveying, please be aware of you are going to ask right visitor, right question. As you know using leading questions leads many problems.

3. Developing Hypothesis

When you are done with analysis and research process now you can create your a/b test hypothesis. Colin McFarland and Rik Higham developed hypothesis kit.

Their framework suggest to form hypothesis in this form;

Based on [quantitative/qualitative insight], we predict that [product change] will cause [impact].

Since sources are limited, you need to prioritize your hypothesis. We use ICE scoring to find best hypothesis to run first.

I stands for Impact. Is your hypothesis aim to fix a tiny bug or it will effect huge segment of users main problems? 5=Highest impact. 1=Lowest impact.

C stands for Confidence. Is your insight solid and straight forward or do you have doubts about that? 5=Highest confidence. 1=Lowest confidence.

E stands for Ease. How much effort you need to test this hypothesis? 5=Lowest effort. 1=Highest effort.

Hypothesis which has highest ICE score is the one you test first.

4. Developing A/B Test

Wow! Look at what you have completed. You now your user segments, conduct analysis and research. Find out things to test and now you ready to run your first A/B test.

In order to run A/B test you need a trustable tool. We use Google Optimize for our clients A/B testing projects. When you use Google Optimize, your data is related directly to Google Analytics. It is very simple to use and free. Click here to get more information Google Optimize.

Almost every A/B testing tool offers an online editor to prepare your test. The main problem with editor is you can not develop highly customized scenarios and sometimes editor can cause UI problems when you live.

We write simple javascripts to manipulate DOM and make changes based on hypothesis in website. My main recommendation is for you to same 😉

5. Reading Results

After performing your first A/B test, you need analyse result of A/B test. If you use Google Optimize, you are one step ahead.

Google Optimize show final result of your test. Based on your test performance you’ll get one of three possible results.

First one is continue to testing process. There are not enough data. This means you shouldn’t finish your test and wait until you reach enough user in test.

Second one is stop test and try a new hypothesis. This means your hypothesis failed. You need to stop test asap and try a new scenario.

Last one is your test is successful. This is the winner case. You shouldn’t go live with your hypothesis until Google Optimize say it is a winner.

If you have any question, please write down in comments. Also check Ecommerce SEO and Ecommerce Instagram Ads.

Leave Your Comment

Your email address will not be published.*