Landing pages

Using A/B testing to optimize performance of landing pages

Landing pages are bread and butter of marketing campaigns, whether they are email, paid per click, or social marketing campaigns. Even with the best-conceived email or PPC campaigns will perform only as good as your landing page performs. This makes landing page optimization one of the top priorities whenever you start the campaign.

What is a landing page?

A landing page is a self-contained page on your website (or a microsite) where you direct visitors acquired by a specific acquisition media or marketing campaign. The reason for using landing pages in ecommerce is that a landing page provides relevant and detailed information about the offer you are making in your acquisition campaign. It makes it easier for a prospect to convert by eliminating any distraction and focusing their attention only to the offer you are promoting.

An example of a landing page

This is the reason why landing pages often perform much better than other pages on your website. An effective landing page consists of several important elements:

  1. Unique value proposition
  2. Benefits for customer
  3. Trust indicators
  4. Imagery
  5. Call to action

Each of these is important and contributes to the conversion rate of your landing page. Of course, each of these elements is variable and by changing it you can make your landing page perform better.

What is the goal of your landing page?

Landing pages have a different purpose, depending on the type of campaign it is tied to. Generally, the goal of a landing page is to make visitors convert and there are several types of conversions you may want your visitors to make. For example, if you run a lead generation campaign you will want your visitors to give their emails. Another example would be a purchase for an online store landing page with goal of making a sale.

The primary indicator of success of a landing page, therefore, is the ratio of visitors who convert - conversion rate. Higher the conversion rate, the more people will convert and you will have more revenue or more email addresses.

In order to measure the success of your landing page, you need to know how many people come to the page, what they do while there, and how many end up converting. For example, you need to track the time your customers spend on the page, which elements they interact with, etc.

How to estimate the performance of your landing page

Before you change any element of a landing page, you need to know how your landing page is performing. Analyzing the performance of any page on your website requires the use of several tools, such as web analytics, heat mapping and if your landing page contains forms, form tracking tools. 

Analysis of the performance of a landing page helps you understand which part of it needs to be improved or changed. It is a critical part of an improvement process you cannot skip.

Web Analytics

Most digital analytics enable you to observe a number of indicators of web page performance. For example, for landing pages, one of the critical indicators of performance is its bounce rate. Bounce rate shows you the percentage of visitors who came to your page and left without making any interaction with it.

A report on landing pages in web analytics

Since the entire point of a landing page is to get a reaction from a visitor, a high bounce rate indicates its failure in its primary function. The reasons for this failure can range from not providing attractive unique value proposal to lacking a persuasive call to action. 

Another indicator of customer engagement is time spent on the page. If a visitor is interested in what you are offering, they will spend some time on the page, trying to gather enough information to convert. Since you know the content on your website, you can estimate the average amount of time your visitors should need to absorb it.

If the time spent on page is significantly shorter than this time and your conversion rate is relatively low, you may want to make the content clearer or add more information to your page. A longer than expected time spent on the page may mean your visitors cannot find the call to action or the information they want.

Furthermore, you can use the reports on the performance of different types of devices or browser versions to see if your landing page provides the same experience to all the visitors. If one category of your visitors are struggling with customer experience on your website, you will be able to find that out in your analytics.

Heat mapping

Heat mapping is a tool that allows you to understand how various parts of your page are performing the best. This is an invaluable tool in landing page optimization, since using it you will be able to verify the soundness of your design. Observing which part of your page and which elements grab the attention of your customers can help you redesign the page.

The purpose of a heatmap is to visualize the behavior of customers on your page. Using heatmaps it is easier to identify the screen elements that are interesting to your visitors and elicit reactions from them.

An example of a heatmap

Heat mapping also allows you to estimate how long your page should be. If people are only looking at your above the fold part of the page, it would make sense to either make the page below the fold more attractive or to remove it entirely. Scroll mapping is a feature of many heatmapping tools and enables you to analyze how far down the page your visitors see.

You can also identify page elements (such as videos, or images), resonate more with your visitors. Click mapping allows you to see what elements (links, images, videos, calls to actions) your visitors click the most. Knowing this information, you can optimize landing pages by changing, replacing or removing unpopular elements.

Form tracking

Many landing pages feature some sort of form, either to capture leads or collect information about the customers. Tracking the performance of forms is the best way to understand how you can improve the interaction between your visitors and the landing page.

An example of a web form

Form tracking allows you to uncover which part of your form works and which fields cause your visitors to quit. Optimizing forms on a landing page is one of the easiest ways to increase conversions, since a form fields can be an impediment to customers completing conversions.

An example of a report from a dedicated form analytics tool

Asking sensitive and personal information as a part of a landing page conversion process requires a great deal of trust from your visitors. Therefore, your forms should avoid asking such questions unless absolutely necessary. Form tracking enables you to understand which questions your customers do not want to provide an answer to and improve information on your landing page to increase the likelihood they will answer by providing more data or remove these fields altogether if you realize they are not necessary.

Testing the changes

However diligent your research is, any findings you make will in large part be based on averages and statistics, as it is usually inefficient to ask every one of the visitors what they did not like on your page. It follows that any change you want to make in order to improve landing page performance will largely be based on supposition and there are many ways to change the landing page.

An example of a report in a hypothesis testing tool

For example, you can change the layout of the page and rearrange the main elements. However, there are a number of ways you can do this. It is impossible to know in advance whether the change you make will improve the performance of the page.

A/B testing - experimenting with variations of a landing page in a real-time environment. Split testing allows you to observe the reactions of your visitors on any variations and compare them to decide which one is performing better. A/B testing is especially well suited for landing pages, as they offer you a high return on investment - landing pages are located low on your funnel and any improvement there will result in higher conversion rate directly.

For example, if you increase the conversion rate on your landing page from 10% to 15% that is the increase of 50% in the final revenue.

What is A/B testing

A/B testing is a method of setting up different variations of a single page online and directing a proportion of your visitors to it. If you have two variations, you would split an equal part of the traffic that comes to the landing page. By comparing the performance of the two chunks of traffic, you can understand if the changes you have made actually result in improving the conversion rate.

A/B testing uses a statistical method called ‘hypothesis testing’, which allows you to calculate the probability of one assumption resulting in better performance than the other. As with analyzing the performance of a landing page, you cannot get the results of how the changes on your landing page impact its performance. This is why a sample of your visitors is used.

A sample represents a limited subset of all your visitors, large enough to allow you to estimate the better performance of your landing page beyond a reasonable doubt. To learn more about the statistics of A/B testing read this article, as it is beyond the scope of this discussion.

For the purposes of testing landing pages, we will divide landing pages according to their purpose. The first group represents permanent landing pages or the landing pages that will be there for a long period of time. The A/B testing on these pages can go on until you optimize them fully, leaving little opportunity to get better performance.

The other group is landing pages for campaigns that will last for a limited or short amount of time, such as seasonal sales. Given limited time these pages are in operation, you would want to test as many hypotheses and as fast as possible.

There are three main types of A/B tests - A/B/n tests, Multivariate tests (MVT) and split-path testing. Each of the methods has its own specific advantages and disadvantages in terms of landing page testing.

For example, A/B/n testing allows you to test medium scale changes on your landing page, such as variations in a copy, images, forms, and similar variations, that you want to validate individually. Its advantage is that you are able to attribute the improvement in conversion rate to a single factor. However, testing every possible change this way takes a long amount of time and if you are using the landing page for a time-limited campaign, you may want to increase the rate of testing.

Usually, an experiment takes two to three weeks to produce results you can rely on. This means if you want to test every improvement individually, you would quickly amount to several months of a testing program, and this may not be an acceptable timeframe.

Multivariate testing allows you to make several low scale changes at once. For example, you can rearrange the layout of your landing pages, changing the position of different elements. Multivariate tests use a matrix of variations to determine which one perform better and can do it in just one test cycle (two to three weeks), eliminating the need for long test program.

The third type of test, split-path testing allows you to experiment with completely different pages, by directing your visitors to different URLs to determine which landing page design performs the best. Its advantage is that it allows you to decide between different landing pages quickly. However, you need to be ready to further optimize the winning page using two other methods.

Bandit testing

When you are running the tests, for example, a multivariate test, for two weeks, the traditional best practice is to leave the tests running until we are certain the improvement is real beyond a reasonable doubt.

However, with sales campaigns of limited duration, landing pages cannot wait until tests complete. In two weeks, your campaign may be over and any variations, improvements and more conversions will be a moot point. In reality, you would have been sending a large proportion of your traffic to the losing variation, reducing your potential revenue for that much.

Fortunately, there is a method to overcome this limitation of traditional testing. It is called bandit testing, where you stop running the tests that have lower performance, directing the traffic to more successful variants. This way you are able to immediately reap the benefits of better performing variations.

The danger inherent to bandit testing is that you may eliminate a variation that may have turned out to be winner later on. On the other hand, the short time available to conduct a test offsets this issue.

What to test on a landing page

Each landing page represents five intertwined elements and variations in any one of them can improve your conversions by increasing visitor’s trust, answering the questions they may have or simply by having a small psychological impact that proves sufficient for a visitor to make a purchase decision.

Testing Call to action button

You can change a call to action button on your landing page in three major ways - changing copy, color and shape, and overall position. By changing the copy of a call to action, you can increase the willingness of a visitor to convert. Sometimes a slight change in wording can make your offer feel more personal and trustworthy. Changing the tone of a call to action from the third person to first may increase the willingness and the relationship they have with your brand.

A call to action also can be tested for its visibility, in terms of its shape, color or position. These are relatively simple tests, and several examples have shown that they can have a significant effect on your conversion rate. 

This is the type of test to run in combination with other changes on the landing page, as the change itself is too minor to merit a separate experiment. There are several indicators that you need to change something in your call to action, such as low conversion rate, a call to action not visible by customers or located near other distractive elements.

Testing headlines

A headline is usually where your unique value proposition is located at. Changing the headline can increase the trust of your visitors, grab their attention and keep them on the page. Testing a headline is usually done initially when you set up the landing page. The primary aim is to make the headline of your landing page connect to your campaign.

A headline can be tested separately or in combination with other elements, such as copy or an image. One of the best indicators you may need to change your headline is a high bounce rate or short visit duration. If visitors cannot understand the benefits of your products, they will not likely hang about your landing page.

Testing landing page layout and design

When your landing page does not perform adequately and the heat maps of your landing pages that visitors have trouble navigating it and finding appropriate information, you can try rearranging the elements so that they are easier to navigate.

Testing landing page layout is the ideal use case for the multivariate test since it allows you to switch layout in different ways and check for a reaction.


Landing pages are the prime candidates for executing a testing program. The reason for this is that the landing pages are located relatively low in your funnel and any improvement of their performance tends to have an immediate positive effect.

Using A/B testing and multivariate testing on your landing pages can ensure that any changes you make can be compared with the original version to determine which performs better. Prior to doing any testing, you need to ascertain what impedes the performance of your landing page and make sure you are addressing these issues.

Quality research is the best way to devise testing hypotheses for your experiments. Research enables you to understand what you want to achieve with your changes and allows you to create a variation with specific solutions to the problems you spotted.

Join the fastest growing ecommerce marketing platform and start growing your business