Our data scientists spend a lot of time trawling through Real User Monitoring (RUM) data. They’re particularly interested in finding useful correlations that we can turn into automatically generated reports. The aim is to make the most useful information easily discoverable.
One of the strongest correlations they found was between site speed and user engagement.
As a rule, we see more pages per session if the website is fast. As load times get slower, session length tails off. When we plot the two on a graph, the following pattern is typical:
The peak in the above graph represents the optimum average load time for the highest engagement. Of course, what this doesn’t tell us is the reason for the correlation. Perhaps the most likely explanation is simple frustration – the slower the website, the more likely people are to give up earlier in their journey.
However, it’s also possible that it reflects limited windows of time that people are willing to spend on a website. For example, if someone is only happy (consciously or unconsciously) to browse a site for two minutes, they can pack more pages into that window if those pages load quickly.
It’s worth noting that there’s almost always a steep ramp up before hitting the peak in session length. In other words, there are always some very short sessions with very fast load times. This may be partly due to sessions that are cut short by fast-loading error pages.
How to use this information
By showing you the impact of site speed on visitor engagement, RUM can tell you how much traffic you’re losing because of slow load times – and how much faster you need to be to improve.