This is the subhead for the blog post
In this blog post, I’ll take a look at three Google Analytics metrics and dimensions that are crucial for understanding user interaction on a website. While these metrics are very useful, you might want to mark them “handle with care,” as they can be misleading and cause confusion in understanding how users are interacting with your page. While these metrics shouldn’t be considered incorrect, I’ll also take a look at how you can “solve” the confusion by extending Google Analytics.
Average Time on Page
Average Time on Page is a useful metric for understanding how users are engaging with your website, and whether they are actually reading your content. It can be a misleading metric, though, especially if your page has a high exit rate.
To understand why this metric is problematic, it is important to understand how Google Analytics measures time on site. The way Google Analytics measures the time a user spends on a page is by calculating the time differential from when that user landed on the page, to when they visit the next page on the website. The key word here is “next.” If a user doesn’t visit another page (an exit), then Google doesn’t know how long the user spent on the page, and they are assigned a time value of 0. Google then calculates the average time on page from the users who did visit another page.
Let’s take a look at the actual formula Google uses to calculate the metric:
Average Time on Page: Time on Page / (Pageviews – Exits)
To use an example how this formula might lead to confusion, imagine we wanted to know the average amount of pizza eaten in a pizzeria on a particular night. Using the Google Analytics method, we add up how much pizza was eaten, and then divide it by the number of people who ate pizza, minus those who didn’t. That number might be a little misleading as there will be a percentage of people who didn’t eat pizza, but it could be considered generally accurate as most people probably ate pizza. Why else would they be in a pizzeria?
Imagine we adjust our analysis slightly and ask how much clam pizza people ate in the pizzeria. Not many people eat clam pizza, but those who do eat a lot. If we take the average clam pizza consumption for that hardcore group, we’ll get a very skewed perspective on the average popularity of clam pizza amongst the whole pizzeria. In the same way, average time on page can become particularly misleading when coupled with a high bounce rate. If most people who visit the page bounce, but those who don’t spend a long time on the page, you are going to have a confused perspective on how users are engaging with the page.
Bounce rate is one of the most commonly referenced metrics used to understand engagement on a page, and it’s also one of the most misunderstood. It is defined by Google Analytics as:
The percentage of single-page sessions (i.e., session in which the person left the property from the first page).
And is calculated as:
Bounce Rate: Bounces / Sessions
One question that I’m asked a lot is “What is a good bounce rate?” The answer really depends on your business. If you haven’t already, a useful activity is visiting the Benchmarking view of Google Analytics, under the Audience section. This allows you to choose an industry and benchmark your site’s performance by channel, location, and device against a number of different metrics, including bounce rate. Retail sites, with strongly defined funnels, are likely to have lower bounce rates (somewhere in the 20-40% range is a good benchmark), than a content website, which is likely to have slightly higher bounce rate (somewhere in the 40-60% range is a good benchmark).
This leads us to the point of confusion, which is this: is a single page session always necessarily a “bad” thing? For example, say you have a long form content marketing piece about your brand, with the purpose of improving brand awareness. If a user visits that page and spends 20 minutes reading that article, but then doesn’t go to another page on your site, Google Analytics considers that a bounce. Yet that person has obviously spent a lot of time engaging with your content, so how do you measure that impact?
If you want to adjust the way Google Analytics defines bounce rate, in order to account for users reading the page, it is possible to add a timer to the page in order to create an event that defines user engagement on the page, and create an Adjusted Bounce Rate. The new definition of bounce rate can be thought of as:
The percentage of single-page sessions (i.e., session in which the person left the property from the first page) less than x seconds.
Google outlines the implementation on their website here, and it is particularly easy to do if you have a tag management system implemented on your site. This will then adjust the bounce rate metric in your reporting. Do consider this carefully before implementing, because you won’t be able to compare the bounce rate before the change to the bounce rate afterwards as apples to apples.
A quick further note while we are on the subject of bounce rate: there is such a thing as a bounce rate that is too good to be true! If your bounce rate is not adjusted and less than 10%, then it is possible that there is an issue with the implementation of Google Analytics.
Understanding how users flow around your website is really important, particularly where the “leaks” are in the funnel. Sometimes, however, it is necessary to link to URLs beyond your own site, and that can cause blind spots in understanding user flow. Unlike referral information, which is automatically stored in Google Analytics, if a user clicks a link to a different site, that information isn’t automatically captured.
Sometimes users of Google Analytics are confused that the next page path dimension doesn’t capture pages linked to beyond the site. Google Analytics can’t record activity on another company’s website, even if Google Analytics is implemented there too. If a funnel page has a mix of links to pages on and off the site, that can lead to a patchwork picture of where users are going.
In order to get a complete picture of what users are clicking on, it is possible to set up an event that tracks all the links clicked on the page, no matter where they go, called an Outbound event. Google outlines the implementation on their website here. This uses a link click tracker to track all the elements on the page and then passes an event to Google Analytics. The event can be set to track the following variables:
Event Category: Outbound
Event Action: click text
Event Label: click URL
So for example, if you linked to an article on the New York Times, your event might look like:
Event Category: Outbound
Event Action: read more here
Event Label: www.nytimes.com/article
That event information can then be utilized to get a full picture of what users are clicking on the page and can be useful for uncovering hidden insights. For example, one page I worked on had an accidentally linked image that was getting a lot of clicks and leading users away from the primary CTA of the funnel. By unlinking the image, we were able to refocus users on the funnel and stop the distraction.
A Final Note
Do note that part of the reason that Google Analytics doesn’t automatically implement these solutions is that they can create a lot of hits. If you are using the free version of Google Analytics and implement any of these solutions, you should be mindful of the data limits within Google Analytics.