Chapter 12 A/B Testing: An Actionable Step-By-Step Guide For CRO

Along your journey of seeking conversion rate solutions, you must have heard about A/B testing. In an online world where we sell to people we’ve never and maybe will never meet, where all interactions and transactions take place online, data suddenly possess tremendous power. 

We human beings are gradually losing belief in our instinctive response. Only data is granted the privilege to back our business decisions, making A/B testing is an almost indispensable part of modern businesses: We test everything and base merely on data to make decisions.

That’s the game. And coming along to this post virtually means you are in with it. Then let’s take a closer look to see how to master the game and what you can do to optimize it.

In this post

I. A/B testing in CRO: What and Why?

A brief definition

Perhaps anyone can tell just by the name: A/B testing is the practice of putting something with two different versions in the same environment, of which the ultimate goal is finding the version with the best performance. 

When it comes to conversion rate optimization (CRO), A/B testing usually means testing factors that affect the conversion rate of a website: Copy, header, images, design, and layout, just to name a few. Each factor has considerable impacts on the conversion rate of a website, and testing to find the best version of the elements contributes significantly to the optimization of the conversion rate.

II. Implement A/B testing on your website

01. Some theoretical steps to take

Regardless of which platform you are using or which object you are testing, any A/B test in the world follows similar patterns:

1.1. Defining goals

To put it simply, a goal is what you want to achieve that leads you to the decision of conducting a test. Just like doing anything in life, you need an explicit goal to back the later testing decisions.

1.2. Selecting what to test

When it comes to optimizing your website’s conversion rate, your options can be grouped into 3 major types based on your goal:

  • Element testing: Individual elements within a page. E.g.: Headline, buttons, copy, etc.
  • Page testing: Combinations of many elements. E.g.: Page layouts.
  • Visitor flow testing: The action you expected visitors to take upon entering your website. E.g.: Multi-step checkout vs Single-step checkout.

Based on research and practical experiences, you decide what to test to improve your website conversion rate.

1.3. Hypothesizing

It’s a super brief plan for the test within one sentence. A proper hypothesis provides you with a decent overview of what would be changed, what’s the impact of the changes and the mechanism behind the impact.

A hypothesis typically looks like this: 

If the add-to-cart button is sticky on the screen regardless of the scrolling behavior, then the conversion rate will increase by 20% because it’s easier for visitors to add products to their carts.”

The number 20% is an indispensable part of the hypothesis as it represents the outcome expected at the end of the test. This is to some extent similar to the KPI of a business plan, and you define the number based on researching or on your personal experience of the previous test.

Note: The 20% figure should NOT be made up based on your personal assumption. Carefully research related cases to earn a practical insight into the change and its possible impacts.

1.4. Running the test

The result of hypothesizing is expected to be different variations of the testing objects. Once the variations are ready, you launch these variations to exactly the same environment to start the test. 

In economics, this unchanged environment is defined as “ceteris paribus”, a term referring to the consistency of all factors in the same scope with the test. The purpose of this is to ensure the result delivered is 100% from the difference of the variations and not from any other factors.

e.g.: If you want to test the change of a headline, then featured image, page layout, text color, font, size, and copy must remain intact.

1.5. Evaluating and Concluding

Looking at data and claiming the winner is not a big deal. Bonus lessons coming along with the test result is a resource you shouldn’t underestimate. Studying data might give you valuable insight into customer needs. 

e.g: If repositioning/adding a call-to-action at the bottom of the page increases the conversion rate significantly, maybe it’s a hint that your above-the-fold design needs improvements.

02. Content Experiments: Begin your A/B test with Google’s free tool

Among many A/B testing tools available in the online market, Content Experiment is a free option allowing you to conduct tests on two or more variations. Setting up an experiment is technically basic enough for fresh beginners to follow. Of course, understanding of HTML and Javascript are still needed in some phases. However, as a free tool with such minimal technical work involved, Content Experiments is not a bad choice, to begin with.

In order for you to easily understand the steps from forming a hypothesis to setting up an experiment, let’s dig deeper into an example to see how steps are processed.

This is a collection page named Cleansers & Exfoliators, displaying cleansers and exfoliators products. You want to increase the conversion rate on this page and decided to conduct an A/B testing. 

So here are the steps you are supposed to take from the moment you come up with the idea to when you successfully set up the test in Content Experiments.

Step 1: Defining your business goal

Your business goal is the ultimate result you want to achieve when deciding to run a test. In this case, your business goal would be:

Increasing conversion page in Cleansers & Exfoliators collection page.

Step 2: Select what to test

After studying this page and a few well-performing pages, you decided there should be a call-to-action on the above the fold area so that customers can easily navigate through the collection.

ab-testing

Step 3: Hypothesizing

Side note: Beside guiding you through the right direction, a strong hypothesis can also serve as your basis to persuade any other people about the necessity for the test.

Your hypothesis:

Adding a “SHOP NOW” button on the above the fold area of the page will increase the conversion rate by at least 20% because visitors can access the product easily just by clicking the button.

Step 4: Configuring the experiments

This is where Google’s Content Experiments comes into play. In order to start the experiments, there are a few things for you to prepare:

  • Different versions of the web page you want to test
  • Google Analytics goals
  • Google Analytics’s Ecommerce tracking enabled
4.1. Different versions of the web pages you want to test

Back to the test, here are the two versions of the Collection page you created based on the hypothesis.

Note: Some A/B testing tools offer users with a page editing feature that allows you to create and edit the testing variation(s) from the original page being tested. With Content Experiments, you create and edit the variation page(s) using your own page builder and/or with the help of your technical specialist.

ab-testing

ab-testing

Another thing to keep in mind is that the two pages must both be live and have unique URLs.

4.2. Setup Google Analytics Goal

Maybe you already know, in Google Analytics, a goal is a set of actions that you expect website visitors to take. In order for Content Experiments to test the performance of the two testing variation pages, you need to set up Google Analytics goals so that relevant actions by visitors are all tracked.

In this case, let’s set up a goal of people clicking on any product in the Collection page.

People visiting the variation page 1 (with no button) will have to scroll down to the Collection section to view the products, while people visiting the variation page 2 (With a “SHOP NOW” button) can click the button and be redirected to the Collection section of the page.

To set up the Google Analytics goal, in your Google Analytics Home, go to Admin.

ab-testing

In the Admin page, make sure you selected the right Account, Property, and View.

(New to these terms? Read Google’s support article about its hierarchy. Haven’t got an Analytics account? Read this guide to Get started with Analytics)

Then click on Goal to go to Goal management page.

ab-testing

Click the New Goal button to create a new goal.

ab-testing

In the Goal setup page, follow the default instruction and provide the required information. Depending on your experience with Google Analytics, you can select either the Template or Custom goal. If the goal can be created with pre-filled configuration, then you can feel comfortable selecting Template.

In this example, we will create a Template goal.

ab-testing

Go on for the next goal settings.

ab-testing

Name: Create a name that reflects the majority of what the goal does so that you can easily find the goal for later use.

Goal slot ID: This is for organizing and grouping the goals when necessary. (Read more about Goal ID here)

Type: Based on the goal to select the proper goal type. 

In this example, we are creating a goal of people who click on a random product within the page and are redirected to the product pages. Let’s track this action by creating a goal of people who visit any product page.

As you can see, almost all product page URLs contain /products/. So we are going to use this character string to tell Google Analytics about the destination page.

ab-testing

By selecting Regular expression with the value of /products/, you are telling Google Analytics that this goal is tracking people who visit any product pages, which is equal to the action of clicking on the products).

ab-testing

Verify the goal to see how often the goal has been converted in the past.

After all the settings are done, click Save and the goal is ready to be used.

4.3. How to enable eCommerce data in Google Analytics

In order to view Content Experiments report data in your Analytics reports, you need to:

  • Enable Ecommerce for the view you want to see the report.
  • Add tracking code to your website to collect Ecommerce data.

Enabling Ecommerce in your Analytics account can be done easily. However, enabling Ecommerce only makes sense when you have the tracking code installed on your website. This task requires a decent understanding of HTML and Javascript. So if you are comfortable editing HTML and coding in Javascript, do follow the official guidelines by Google Analytics about Setting up Ecommerce Tracking. In case you have any difficulty taking these actions, asking for the help of a technical specialist would be recommended.

By the end of this step, you should be ready for Content Experiments with the required elements prepared.

4.4. Setup your experiments

When the required elements for an experiment are ready, from your Google Analytics Admin, go to Behavior > Experiments > Create Experiment to create a new experiment.

Fill in with the required information to set up the experiment.

Name for this experiment: Give the experiment a name that best reflects the experience for later identifying.

Objective for this experiment: Select the goal created in the previous step.

ab-testing

Percentage of traffic to experiment: Ideally, 100% of traffic is recommended for a quick experiment result.

Fill in with the URLs to the two variation pages.

ab-testing

In the last step of the configuration, Content Experiments offer the tracking codes to be added into the variation pages. 

Once again, if you are comfortable with editing HTML and coding in Javascript, then do follow Google’s instructions on installing Google Analytics tracking code and adding script code to the variation pages. If the code confuses you by any means, contact a technical specialist to avoid any unwanted accidents.

ab-testing

Once the code are properly installed where necessary, go ahead and launch the experiments. 

IV. Conversion Rate Optimization: What to test

If you are confused with the question of what to test to improve and optimize the conversion rate on your website, keep this in mind: The factors with the most impact on the conversion rate are the first factors to be tested.

As your time and budget are limited, carefully consider the testing objective based on the nature of your website and the impact of each element on the conversion rate.

01. Headlines

A headline is the first thing visitors see when entering a web page. It plays an important role of intriguing visitors and keeps them navigating. Headlines have a significant impact on the session duration or time-on-site and therefore considerably affect your conversion rate.

When generating headline testing variations, keep in mind the following criteria for a great headline as follows:

  • Keep them short and precise
  • Highlight customer benefits in the first words
  • Adding statistics and numbers makes the headline more compelling
  • Make it stand out from the background color.

02. Copy

A copy is anything within your page that delivers a certain message to your target audience. Based on the copy’s purpose, there are different criteria for a good copy. However, there are general standards that apply to all types of copy in order to improve the conversion rate.

  • Focus on customers, not brands.
  • Offer solutions for customers’ problems, highlight the benefits they receive.
  • Easy to read, follow and comprehend.

03. Design and layout

Design and layout refer to the combination of many or all elements within a page. These two factors affect the overall experience of visitors to your website so the impact they have on the conversion rate is enormous. 

Testing the design and layout of a website might be a lot more complex than doing the same thing with a single element. But optimized improvement in design and layout usually delivers a multiple-times bigger increase in the conversion rate.

To create an optimized design and layout testing variation, carefully study user behaviors to identify the common navigating paths. The best web design is the one that best facilitates the sites navigation.

Note: Keep an eye on the responsiveness for different devices. Device diversity is manipulating the way people behave in the online world and you’d better synchronize yourself with it.

04. Graphic and images

Graphic and images are used to complement copy so that messages are delivered more thoroughly to visitors. A website with smart graphics and good images at least leave visitors with a better impression. Improving these elements is a way to improve your conversion rate.

Keep in mind:

  • Images MUST always be high in resolution.
  • Images should never be just for decoration. Use images as a supportive part of your content.
  • Funny Infographics draw visitors attention effectively and easily.

05. Forms

A form is a special element that asks for people’s interaction with the website, or more specifically, leaving their information. Asking for information might frustrate your visitors easily, so forms have a direct impact on the conversion rate.

If you plan on testing a form, researching on which fields your potential customers abandoned the form could give you valuable insight into what information to ask in the form, and whatnot.

Tips: The tool to track where visitors abandon the form ACTUALLY exists. It is called Field Bottlenecks. Go ahead and try it to see if a smarter form increases your conversion rate.

06. Call-to-action

Of course, call-to-action is an important touchpoint. Most conversions take place after people click on a call-to-action. Optimizing call-to-action is a direct way to optimize the conversion rate.

Depending on the website’s style, make the call-to-action as noticeable as possible. Also make sure the call-to-action is within your visitors hand-reach. If the site’s content is long and contains many parts, consider adding call-to-actions to every section of the page. Be available whenever there’s a potential conversion and get ready for it.

V. A/B testing Conditions

You’ve heard people talking about conversion rate optimization and A/B testing everywhere. You are eager to start your own A/B testing and optimize your own website’s conversion rates. However, there’s a fact that A/B testing is not for everyone. And conducting A/B tests while essential conditions are missing is totally a waste of time and effort.

So, before diving in and planning an intensive A/B testing campaign, take a look at the following factors to re-consider carefully whether and when you should start.

01. Your monthly traffic/conversion amount must be big enough

As a type of statistical hypothesis test, A/B test result needs conversion volume to gain statistical significance. Many CRO experts suggest you should only think of A/B tests when your conversion volume exceeds 1000 conversions per month. Well, for many startups and micro businesses, that threshold is almost equal to “NEVER” (Yes, i’s an exaggerated expression). 

That is just to say, your conversion volume must be big enough. 1000 monthly conversions probably sounds a bit strict. But if your conversion volume takes half a year to reach this threshold, you’d better put off A/B testing and try other Conversion Rate Optimization techniques.

02. Only test when you need more proof to back your decisions

Many businesses obsessively apply A/B tests to every decision with the belief that only data can say for sure. Theoretically, this is true. However, keep in mind that setting up and running A/B tests cost you not just time and money. Also there are the effort and human resources spent on designing and monitoring the whole testing process. Spending time on testing things that you almost know for sure is totally a waste.

If a version is widely accepted by common sense and you are pretty sure it is the better version, then be brave and skip the A/B tests.

03. You testing decision must be backed by a strong hypothesis

Hypothesis is not just a statement, it is the embodiment of a strong need. A strong hypothesis reflects the necessity and urgency of the change. 

You think an element should be tested but you cannot form a strong hypothesis, no matter how hard you try? Maybe waiting until you’re able to create one is a good solution. Testing random ideas is just a kind of temptation that is costly and time-consuming. Keep calm and resist it.

Congratulations! You’ve just finished reading the guideline to the A/B testing game.

Good luck on your journey of conquering website conversion rates.

Try out all PageFly features with Free plan

.myshopify.com
Don’t have Shopify store yet? Create store
×