Do Banners Really Work?

Published in Online Marketing.

Watch our video

There is a better version of your web

Share this post

After talking about heatmaps, let’s proceed with another controversial topic related to conversion optimization: using banners in websites.

Online advertisement has evolved over the years. If you are an experienced web surfer, surely you remember those web pages in the 90s with lots of banners and animated GIFs dancing inside and trying to get your attention so you end up clicking on them.

But it’s 2020. Banners have changed (in addition to many other things). And what you should ask yourself is: Do web banners really work today? Let’s discover it with a simple example that you can also apply to your website.

Looking For The Greatest Impact

The key point of banners is that they try to catch the attention of your visitors so that they click on them and thus visitors end up visiting or doing where/what you want. This is why you have to add banners where they can potentially have more impact.

A good place to add a banner is in those contents that get the largest amount of visits. To know this information, you just have to take a look at the analytics of your website. We did it and here are the results:

Most visited content on the Nelio Software website.
Most visited content on the Nelio Software website.

As you see, the most visited content on our website are two blog posts, one in Spanish and one in English. In the first we describe common problems that you may have in WordPress, and in the second we explain how we migrated our website and what problems you can find if you do the same (it looks like talking about problems gets visits… Therefore, if you have not done so, it is a good topic to discuss on your website. See? Apart from nice examples to test on your website I also give you ideas of content that may work for you! 😉)

Back to the test, our goal is to drive the traffic that comes to visit the two most popular posts on our blog to the pages of our products. And for this we will use banners. Specifically, we are going to use a call-to-action block that we have developed in Nelio for Gutenberg, the WordPress block editor.

This block will be included within the content of the most visited posts to see if visitors click on it and end up visiting our product pages. To see if the banners really work, let’s test the original version of our blog post without banners against a variant that does include banners. Thus, with a simple A/B test we can see results and draw conclusions.

Creating A/B Tests to Understand The Way Banners Work On Our Site

Let’s create two A/B tests of posts to test the banners on our top posts, both in English and Spanish. We will see how our audience behaves and if the banners are really useful or not to drive traffic to the pages of our products.

We use Nelio A/B Testing and create a new test of posts. Then, we complete the necessary information to create the test as you can see in the following screenshot:

Page to create a new test of posts with the configuration we want for our A/B test in the Spanish site.
Page to create a new test of posts with the configuration we want for our A/B test in the Spanish site.

We have defined a title and a description of the test and we have selected the post to be tested. In addition, we have created a variant post with the banners. Finally, we define as conversion action the visit to the Nelio Content page, which is the product to which we will direct visitors from the Spanish site through the banners in the variant.

This whole process is very fast. Having the test created and configured is extremely simple. In addition, editing the variant of the post to add the banners is also easy, since we only add banner blocks using WordPress’ built-in editor. They could be image blocks, but as I anticipated before, we will use a special block type created by us that will work as a banner.

The result of the post in Spanish with the banner inside the content looks like the one you can see below:

Fragment of the variant including banners for the Spanish post.
Fragment of the variant including banners for the Spanish post.

In fact, we have added 4 banners along the content of said post with different colors and texts to get the attention of visitors who end up seeing the alternative version of the original post.

For the English version, we do the same but here we will direct visitors to Nelio A/B Testing, our second product. The configuration is as follows:

Page to create the A/B test for the post in English. Here the goal is to visit the Nelio A/B Testing product page.
Page to create the A/B test for the post in English. Here the goal is to visit the Nelio A/B Testing product page.

For the sake of transparency, here you can see how both variants looked like. Please keep in mind these are very long screenshots… that’s why I link them here instead of embedding them in the article:

Now that we have all the ingredients ready, we just have to start each of the A/B tests so that they begin to track and collect the behavior data from our visitors.

Analyzing The Results: Did The Banners Work?

Before answering this question, let me give you some additional information about the A/B tests performed…

Each test was running during 11 days and collected the behavior of almost 3,500 visitors, which is a pretty interesting sample of our total audience. During all this time all we have had to do is wait. When you do A/B testing, all the work is done at the beginning, when configuring the tests and the variants to be tested (which as you saw was not that much work). Then you only need to relax and wait for the results to arrive.

In this case, let me anticipate that the results are inconclusive. We have not found a clear winning variant in either of the two A/B tests. Let’s look at the results of the A/B test of the Spanish post:

Even though the variant with banners (B) is somewhat worse, the results are not statistically strong, so a clear conclusion cannot be drawn.
Even though the variant with banners (B) is somewhat worse, the results are not statistically strong, so a clear conclusion cannot be drawn.

As you can see in the previous graphs, both the original version (A) and the variant with banners (B) of the A/B test have similar results.

It seems that the version with banners is somewhat worse in terms of the conversion rate (that is, the number of visitors who click on a banner divided by the total number of visitors). However, we cannot only look at this isolated data. We need to check if the results are statistically significant, and in this case they are not.

Nelio A/B Testing gives us this information and indicates that with these data it is not possible to decide if one variant is better than the other. Thus, we have the possibility to stop the test and try another type of change or be patient and give it more time to see if, as time progresses and we get more data, anything changes.

Now, let’s look at the results of the A/B test of the English post:

The variant with banners (B) has a much better conversion rate than the original version of the English post. However, even with a 220% improvement, the number of conversions is so low that the results are not statistically significant.
The variant with banners (B) has a much better conversion rate than the original version of the English post. However, even with a 220% improvement, the number of conversions is so low that the results are not statistically significant.

Looking at the previous graphs one can think that it is clear that the version with banners is much better than the current version of the post. The conversion rate of the version with banners is 220% better, so we should choose this variant as the winner of the A/B test, right?

The truth is that here happens the same as with the A/B test of the Spanish post. The number of conversions received in this test is very low. Only 4 visitors ended up visiting the Nelio A/B Testing page, and of these only 3 did so through the banners of variant B.

With such a low number of conversions with respect to the total number of visitors received, it is impossible to state that one variant is better than the other.

Banner Blindness is Real

After analyzing the results we shouldn’t think that we wasted time. In fact, we have learned many things. Adding banners on top posts didn’t work on our website and for our specific audience.

In addition, if we go one step further and explore the heatmaps our A/B tests provide, we realize we’re in front of a clear case of banner blindness:

The banners used in the variants are so obvious that the visitor ignores them directly. Therefore, it is not worth leaving the two A/B tests running any longer to wait to see if the results change. They will not.

In this case it is smarter to think of testing other types of ads, much more integrated with the content of each post itself. I am thinking of using links to our products within the content, as long as this makes sense.

Or we could even use Google’s technique of camouflaging banners so that they seem part of the content, just as this search engine does with the ads in the search results. Therefore, we should be much more creative with the design of the banners.

The conclusion that we draw from all this is that the typical kind of banner we all know is dead. Visitors are able to detect and ignore them, so you need to test other options to get their attention.

But remember that we do not intend to be the greatest authority here with the results of the A/B tests and the conclusions drawn from them. Banners within the content didn’t work for us, but this does not mean that they will not work for you. You should test it yourself on your website. After that, don’t forget to let us know what you discovered.

Featured image by Andrew Neel on Unsplash.

Leave a Reply

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

I have read and agree to the Nelio Software Privacy Policy

Your personal data will be located on SiteGround and will be treated by Nelio Software with the sole purpose of publishing this comment here. The legitimation is carried out through your express consent. Contact us to access, rectify, limit, or delete your data.