The dreaded 100 percent bounce rate and what it’s really saying about your content
Have you ever encountered the dreaded 100 percent bounce rate in Google Analytics, and wondered what it’s really saying about your content? The short answer: absolutely nothing at all. The long answer, on the other hand, requires a deeper dive into bounce rates. Mostly, how Google Analytics (GA) calculates metrics such as session duration (time on site) and time on page—which publishers are too often ready to accept as a telltale indicator of their content’s performance.
While it makes sense to turn to session duration and time on page when determining how engaged your visitors are, these metrics are accompanied by a far less reliable measure—bounce rate.
The problem with bounce rates
Yes, the dreaded bounce rate—bringing fear into the hearts of marketers since the early days of the internet. For the most part, we’ve probably all been taught that a high bounce rate is indicative of poor quality content; but, that’s just not the case.
Based on GA’s definition, “bounce rate” is the percentage of single-page sessions. This implies that a visitor could land on your page, consume all of your content, exit off the same page and still count as a “bounce”—for the sole reason of not advancing to a second page. (This, of course, is a simplistic way of interpreting bounce rate—we’ll get into the specifics concerning measurement later—but it’s important to remember.)
Not all “bounced” sessions are created equal. A five-minute session obviously signals higher visitor engagement than a 30-second session, but this time distinction may not be reflected in the data GA displays (assuming these were single-page sessions). Translation: this is bad news for content-heavy sites that are only producing content as its main attraction.
Now that we know the flaw in bounce rates, we can get into the important part: how bounce rates affect GA’s time measurements like session duration and time on page.
GA will calculate session duration (time on site) in one of two ways, depending on how it was implemented. In the case where no event tracking is implemented, GA will calculate session duration to be the difference between the timestamps of the first hit on the last page a user visits, and the first hit of the first page that a user lands on.
For example, if a visitor lands on your page at 11 a.m., and clicks out to another page on your site at 11:03 a.m., GA would determine that session duration to be three minutes. However, this calculation does not take into account how long the same visitor spent on the second page if they failed to move on to a third page.
In the case where event tracking was set up to account for on-page engagement hits—such as when a visitor starts scrolling, hits “play” on a video embed, or gets to the bottom of your page—GA will now calculate session duration to be the difference between the last engagement hit fired on the last page, and the first hit on the first page that a visitor lands on.
If we were to apply this tracking model to our previous example, and say that an engagement hit on page two was fired at 11:04 a.m., before the visitor exited off the same page, then the session duration would have been calculated to be four minutes instead of three—a more accurate reflection of the total time the visitor spent engaging with your site.
Where no event tracking was implemented in the first scenario, if a single-page session were to occur, GA will have no way of knowing how much time was spent on that page as no subsequent actions were taken by the visitor. Since the difference between the timestamps between hits are used to calculate session duration, a single-page session, in this case, would yield a 0:00 session duration by default. When examining your site performance, you may have noticed that every page with a 100 percent bounce rate also has an accompanying session duration of 0:00.
When calculating time on page (rather than on site—of which a “session” would classify as), the same calculations as session duration apply. The only difference is that the time displayed would represent how long a visitor spent on a single page, rather than their session as a whole.
Up until this point, we’ve examined how session duration and time on page are calculated as single visits. However, GA will show these metrics as averages instead, which leads to some confusion when interpreting the data.
Average session duration is calculated by dividing the total session duration during a specified period, by the total number of registered sessions during that same period. What this means is that all “bounced,” single-page sessions with a session duration of 0:00 are also included in the calculation, effectively bringing the average down. Therefore, your average session duration will most likely be much lower than what it is in actuality if your site’s bounce rate is high.
On the other hand, average time on page does not take into account any bounces or single-page sessions. Instead, only sessions logging more than one page visited per session are used to calculate the average—explaining why the average time on page is usually much higher than your average session duration when your site’s bounce rate is high.
Unfortunately, this also means that this metric may only provide insight in regards to a small portion of your overall site traffic. It cannot be used in an accurate, holistic interpretation. (For example, if your page’s bounce rate is 95 percent, then the average time on page would only be looking at the 5 percent of visitors that made it to a second page).
Armed with this newfound understanding of bounce rates and how engagement time is measured in Google Analytics, what can you do to measure your content’s performance more accurately ?
1. First and foremost, implement some form of custom event tracking to fire engagement hits, even in the case of a single-page session.
This will improve not only your bounce rate, but also your average session duration and average time on site. Because visitors that fire an engagement hit are no longer considered a “bounce,” your bounce rate will decrease, even if the number of single-page sessions do not. Time metrics like average session duration and time on page should also increase, as GA’s calculations now shift to scenario two, which we outlined above. (Quick recap: single-page sessions will now yield subsequent timestamps, as engagement hits are triggered while a visitor browses through your page.) This article is a great starting point for implementing better content performance tracking measures.
2. Even without implementing any additional code, you can better gauge your content’s actual performance by using tools with built-in analytics and measurement capabilities to corroborate any data GA provides.
For example, Quietly is able to measure a variety of engagement metrics—such as time spent on page, which parts of your article is exciting readers the most, and drop-off rate, just to name a few—for content that is created using our proprietary tools.
3. Finally, you can also improve your site’s bounce rate by making sure that your content can be easily discovered.
Incorporate calls to action within your articles, and recommend more content at the bottom of a page, or in your side navigation. Providing quality content is useless if no one can discover it, so make that a priority when updating your site.
Image: Erik Schepers/Flickr