More scroll depths with website A/B tests via Google Optimize

Improve scroll depth with website A/B testing via Google Optimize

In addition to increasing the conversion rate and reducing the bounce rate, we can use Google Optimize to perform other exciting website A/B tests. For example, tests to determine a particularly binding website version that encourages users to read the landing page to the end if possible.

Landing pages, which are primarily designed for lead generation, often have a tried and tested standard structure: They start with an exciting head area, appealing headlines, then move on to an information area in the middle and end with a contact form that the website user can use can register for an offer and thus enter a funnel.

The goal of such conversion landing pages is always to maximize the conversion rate. That means: What percentage of a hundred users become leads and make use of our offer? This article explains how a conversion rate can be increased directly in a contact form or on an order page with A/B tests. This article says that users stay on the website at all, instead of jumping off 90% of the time.

Now, these two articles will focus on the first and last parts of scrolling in a landing page layout. In between, a structure section is missing, namely the middle part – the content block – through which users have to scroll to get to the contact form. To do this, we can use a scroll depth test to find out which content most strongly encourages users to continue reading, instead of leaving the landing page halfway when scrolling.

Classification: How relevant is an improved scroll depth for the conversion rate?

An increased scroll depth ensures that more users consume the content at all. Where users do not consume the content and leave the landing page, they accordingly cannot even get to fill out or even see a contact form.

With a general bounce rate of 90% and a low average scroll depth, where only 20% of the remaining 10% as a subset make it to the contact form, the calculation would be: 100 users arrive at the landing page, for example via Google Ads with one CPC (Cost-per-Click) of, for example, 2 euros – that would be an advertising budget of 200 euros. With a 90% bounce rate, only 10% remain, meaning that out of those 100 users, effectively only 10 users stay on the site at all.

The cost point for a user who consumes the content at all would therefore be 20 euros per user at this level. But it goes even further: With regard to scroll depths, only 20 percent of this subset comes down to the lowest section – the contact form. That would be 2 users out of 100 who make it to the bottom.

In the end, in this realistic example, only 2 out of 100 users see the contact form and even consider sending it. Users need to consume the content before the contact form to understand what this offer is about and to be convinced to submit the request at the bottom of the landing page.

Save scroll depths in Google Analytics 3 and Google Analytics 4 as metric

Since Google Optimize works with Google Analytics in the measurement, a scroll depth measurement must first be recorded there. In Google Analytics 3 and Google Analytics 4, percentage sections for scroll depths must be stored in Google Tag Manager as an event with the Scroll Depth trigger and output as an event to Google Analytics 3 with a Google Analytics: Universal Analytics tag. When configuring the trigger scroll depth (Scroll Depth), you can optionally specify a division of the landing page scroll depth in percentage sections: We recommend division in 25 percent increments or in 10 percent increments.

Implementation of scroll depths as a trigger for a Google Analytics event in Google Tag Manager.

However, Google Analytics has limited data processing capabilities when collecting events, so splitting into 10 percent increments could quickly overwhelm the data collection in the account. Ergo: In the event report, instead of a green shield icon, there could be a yellow shield icon at the top, which indicates, for example, that the data is only available for 80% of the sessions and not for 100%, since the data volume of the event recording is the volume of a free Google -Analytics account blows up.

Therefore, for accurate and safe data collection, the 25 percent increments would make more sense when we are dealing with very high traffic sites, because each scroll that shows up as a counter in the percentage section counts as an event. With 10 percent increments, we would have 10 events or more per user – if they scroll up and down. With 25 percent increments, that would be 4 events per user – or more or less.

If the event counting capacity is overloaded, in the normal version of Google Analytics 3 it can be indicated that the given data is no longer based on 100% of the sessions, but on a smaller proportion. This can be the case when so many events flow into the system through the scroll depth measurement in the form of 5 percent or 10 percent increments per user and session that it automatically stops counting many events.

Set scroll depth as goal metric for measurement in Google Optimize

As soon as the scroll depth is output as an event to Google Analytics 3 or Google Analytics 4, we can select this in Google Optimize as the target metric of the A/B test.

What do we get from website A/B testing on scroll depths?

We use scroll depth tests to see the percentage of users who reach a specific section of the landing page. Most people’s website user behavior is extremely fast-paced, and we can always assume that the percentage of lingering users will decrease with each landing page section level. Let’s assume a division into 25 percent increments: In this case, the first 25 percent is the top container and the last 25 percent is the bottom. As a rule, far less than half of the users will reach the bottom container. What interests us are the middle two levels, because they indicate how the middle blocks of content affected exit behavior.

Now let’s move on to the testing strategies.

Test 1 - Content Blocks: Information Hierarchy

What interests users first? Is it worth getting straight to the point, or is it more of a substantive introduction than a bridge? We can only estimate this in advance based on experience as a guideline. In practice, we can only determine which content block extrapolated to 1,000 but really arouses the most interest to continue reading with an A/B website test. With Google Optimize, content blocks can be moved and arranged in the front-end editor using drag-and-drop.

Test 2 - Content blocks: Content and visual information design

The content and visual design of the information is once again about testing headings, images, texts and buttons (if there are any). The tests should always be run one after the other, ie: Only start a new test when the previous one has been completed, so as not to mix up the results. Here it says again, as in the instructions for reducing the bounce rate and increasing the conversion rate: test the sizes, contents and arrangement of the individual content blocks.


  • Test the size of the headings
  • Test the color, font, and style of the headings
  • Test headline content: Try convincing approaches, perhaps with the help of copywriters or best practices from professional copywriters

Test images:

  • Test image sizes
  • Test image content

Test texts:

  • Test text content
  • Test text element highlighting: Print important passages in bold or as a quote in colour, larger and highlighted

Test 3 - Equip content blocks with navigation elements: tables of contents

Sometimes the overall scroll depth can be increased to 1,000 by implementing a table of contents. This may not sound that spectacular, but it is actually the classic in scroll depth improvement. It seems important for many users to have at least a rough idea of the length and content of the landing page.

Such a table of contents can be a conservative-looking list at the beginning of a text, a small, smart list on the left or right sidebar, or a small box that moves with you when you scroll – on the side or in the header bar. Ultimately, the motto here is again: test, test, test.

You might also be interested in:

Reduce bounces with website A/B testing via Google Optimize

What improvements need to be made to the website to reduce the bounce rate?

Increase conversion rates with website A/B testing via Google Optimize

If we want to find out which website version gives the best conversion rate, we can do that with an A/B test.