In our previous report entitled Unfolding the Fold we analyzed web visitors scrolling behavior and determined the effect of the so-called ‘fold’ on browsing activity.
One year later we are excited to publish much more detailed research, revealing new information and providing you with a whole new set of best practices and recommendations. This research is based on a sample of a data that was collected by the ClickTale service from thousands of websites and over 80,000 page views chosen at random between June 15th and July 15th, 2007.
What did we learn last time?
- 91% of the page views were long enough to contain scroll-bar. Of those, 76% were scrolled to some extent.
- 22% of the page views with a scroll-bar were scrolled all the way to the bottom.
- Visitors are equally likely to scan the entire page no matter the page size.
- The fold is not a single location, but a broadly dispersed distribution with three peaks located at roughly 430, 600 and 860 pixels. These peaks correspond to the three most popular screen resolutions used today: 800—600, 1024—768 and 1280—1024, minus about 170 pixels used up by the non-client area of the browser. In other words, depending on the screen size, users will typically see at least 430 and up to 860 vertical pixels worth of information on their screens when they open their site, without having to scroll.
We measured visitor scrolling using a new metric called Visibility which is defined as the fraction of page views that scroll to a specific location in the page. The location visitors reach on the page is called Scrolling Reach. When scrolling reach is measured as a relative number, it is expressed as a percentage of the page and when it’s measured as an absolute number, it is expressed in pixels.
Page views were aggregated into 9 page height groups that vary from 2,000 to 10,000 pixels. To qualify into a specific group, a web page had to be within 2-10% of the designated group height. Once a page qualified into a height group, we analyzed the visitor scrolling behavior and extracted insights about their browsing habits.
Note that this research is based on page views, not visitors. Therefore, any given page view doesn’t necessarily reflect the first visit to a page.
- As can be seen from the “Visibility vs. Absolute Scrolling Reach” graph, for content above the 1,000 pixel line, visitor scrolling habits are remarkably similar regardless of page height. This similarity in scrolling behavior is due to the fact that above the 1,000 pixel line Visibility is primarily affected by the distribution of the fold and by non-scrolling visitors.
- For content below the 1,000 pixel line, scrolling behavior seems to diverge in the absolute pixel scale. However…
- Almost identical percentages of page views (15%-20%) reach the page bottom regardless of page height.
- We define the page area Above the Fold as as the area above the 1,000 pixel line covering over 99% of observed fold locations. The similarity in scrolling behavior Above the Fold across the various page heights can be seen more clearly in the chart below:
- When we examine the “Visibility vs. Scrolling Reach” data as expressed in the graph below, exposure rates appear to diverge at the page top and converge at the page bottom.
- At the page bottom, exposure rates are almost identical and range between 15-20%.
- The divergence at the top of the page might be due to page height and the fold effect: The fold represents a larger percentage of the shorter pages and therefore a larger percent of those pages are viewed at close to or at 100%.
- To test this hypothesis, we reran the analysis focusing on scrolling behavior Below the Fold. The results were dramatic.
- It appears that regardless of page height, scrolling reach is very similar on the relative scale with some resemblance to a linear model (or a very flat exponential): between 64% to 68% of the page views are likely to reach the 1K pixel line and 15% to 20% will reach the bottom of the page.
By modeling the relationship between visibility and scrolling Reach below 1,000 pixels, we find a statistically significant linear regression with the following characteristics:
- Visibility = 57.6 – (0.409 x Scrolling Reach)
where Scrolling Reach (from 0 to 100%) is a the fraction of the page scrolled below the 1,000 pixel-line.
- Coefficient of Determination (R-squared) = 89.4% which indicates that the linear model is a very good fit for the scrolling data.
Hence, it appears that visitors scroll in a relative way – about the same percentage of page views will reach the middle of a web page regardless of the actual page height in pixels.
Visitors appear to be using the location of the scroll bar but not the size of the tracker when scrolling, since the scroll bar location is a relative indicator and the scroll tracker size is an indicator of page height.
You can also find out how your website measures up relative to the averages reported above by using the ClickTale Heatmaps and analytics service to record and observe your own visitors’ online behavior. Sign up to measure how visitors scroll in your web site!