What’s this misunderstood metric: bounce rate? Google Analytics provides the option to view it, but what counts as a bounce and why should you care?
It’s often used as a signal of engagement for marketers, something to be lowered at all costs. A high bounce rate for many people means very simply that users don’t like your page; it means there’s some sort of mismatch between your experience and their expectations, and that could indeed be true.
But it also could be a meaningless metric, or maybe even a good thing, depending on the context and how you’re measuring bounce rate.
This guide will outline what bounce rate is actually measuring in Google Analytics, what that number means for your marketing, why it should be analyzed differently in different channels, and how to tell what a good or bad bounce rate actually is.
We’ll also cover how to lower bounce rate, both from a technical perspective and a user experience perspective.
What is Bounce Rate in Google Analytics?
If you asked a random marketer what bounce rate in Google Analytics measures, they’d probably say something like, “the percentage of people who leave after viewing only one page.”
This, however, isn’t totally true. It’s a simplified version of what bounce rate actually measures, and many people misunderstand this (for example, almost every blog post about bounce rate uses this definition).
So, what’s the actual definition of bounce rate? Google Analytic registers a bounce as a single engagement hit.
That means that, for instance, if a user enters a landing page and leaves, they may be recorded as a bounce. But if they enter the same landing page and hit “play” on a video that you have set up as an “interaction” event, and only after that do they leave, they won’t be counted as a bounce.
Both examples are only single page sessions, but only the former would be counted as a bounce. The point is, how you set up your event tracking affects your bounce rate. Google Analytics registers Interaction events as an engagement hit, but they alter your bounce rate.
To further elucidate, let’s walk through a real example.
Let’s say a friend told you about MyFitnessPal, a cool way to track what you eat and your exercise activity.
You enter their homepage, scroll down all the way to the bottom, clicking a video and messing around with a few images and case study modules. You’re clearly interested and interacting with the website. Yet, the site hasn’t set up event tracking for any of these additional interactions. So, when you leave the site without visiting any additional pages, you’re counted as a bounce.
I made a GIf of this type of user behavior, and I also included what registers to the Console when using Google Analytics Debugger, a great tool for auditing your analytics.
This tool will show you when your behavior triggers a pageview or event–and as you can see, only a pageview has been triggered (and an event to say you viewed an ad, even though the ad is at the very bottom of the page).
Now, let’s pretend you’re on the market for a new snowboard.
Burton is a household name, so you hop over to their site–but you get hit with a pop-up right away.
A few different events are being fired immediately. Some are recorded as nonInteraction, because you’re simply viewing the carousel feature. But when you click away from the pop-up, an Interaction event is triggered.
Because of that, even if you exit right after that, your visit is not a bounce.
So, two very similar sessions can be counted as a bounce or a non-bounce, and it all depends on how you’ve defined what a bounce is in your account. In the Burton example, many marketers and analysts will call this a type of “adjusted bounce rate,” which just means that you’re defining bounce rate differently that Google Analytics will out of the box.
Most often, people will simply set up an Interaction event after a user has spent a certain amount of time on their site, but you can make any event you track an “Interaction” event, which will alter your bounce rate.
For instance, a good example of an Interaction event is how Mailshake uses their “demo” video as an indicator of user engagement. As soon as you click the button to play the video, an Interaction event is fired and you’re counted as an “engaged” non-bounce session:
Before we move on, another point of confusion: what’s the difference between bounce rate and exit rate? These two metrics are sometimes used interchangeably without fully understanding the difference.
Here are the definitions:
- Bounce Rate: the percentage of single-engagement sessions
- Exit Rate: the percentage of exits on a page
In other words, an exit rate measures the percentage of exits on a page (the last page in a session) compared to the total number of pageviews. On the other hand, the percentage of exits for sessions that also start on that same page is considered the bounce rate. Google Analytics puts these metrics next to each other on lots of reports, so it helps to keep that in mind.
Here’s a great example taken from the Google Analytics knowledge base. Pretend you have a website with the following user behavior:
- Monday: Page B > Page A > Page C > Exit
- Tuesday: Page B > Exit
- Wednesday: Page A > Page C > Page B > Exit
- Thursday: Page C > Exit
- Friday: Page B > Page C > Page A > Exit
Your exit rates would be:
- Page A: 33% (3 sessions included Page A, 1 session exited from Page A)
- Page B: 50% (4 sessions included Page B, 2 sessions exited from Page B)
- Page C: 50% (4 sessions included Page C, 2 sessions exited from Page C)
And your bounce rates would be:
- Page A: 0% (one session began with Page A–but that was not a single-page session, so it has no Bounce Rate)
- Page B: 33% (Bounce Rate is less than Exit Rate–because 3 sessions started with Page B, with one leading to a bounce)
- Page C: 100% (one session started with Page C, and it lead to a bounce)
Now you should have a good understanding of what bounce rate in Google Analytics is, so let’s move on to one of the most common question regarding bounce rates: what’s a good one?
What’s a Good Bounce Rate in Google Analytics?
What’s a good benchmark to aim for? It’s a common question, and for a good underlying reason: no one wants to be behind the competition, even with a metric like bounce rate.
There’s a post on the HubSpot blog that gives these numbers for average bounce rates for specific types of sites:
- Content Websites: 40-60%
- Lead Generation: 30-50%
- Blogs: 70-98%
- Retail Sites: 20-40%
- Service Sites: 10-30%
- Landing Pages: 70-90%
If I were you, I would ignore those completely and not spend any more time looking up average bounce rates.
We’ve already covered how a bounce rate is defined, and how it’s anything but a universally defined point of truth. But it’s also important to note the complexity and variance of any given website’s traffic makeup, as well as the context of a given page within that website.
For the first point, just take a look at your Google Analytics right now. Go to Behavior > Site Content > Landing Pages and order them by bounce rate (use the comparison feature on the upper right corner). The variance is striking even within one site.
I also exported a CSV file with this site’s bounce rates, excluded all of the pages at 100%, and made a histogram in R. There’s a lot of different bounce rates here (if you’re curious, the mean bounce rate of this dataset is 55.52%):
That’s not to say that your website’s average bounce rate has no value. It’s just to say that there’s a lot of truth hidden in that average. The graphs above show that there are many different bounce rates across the site–but it doesn’t tell you what the purpose of any of the pages is, how many visits it gets, or what the traffic sources are.
So, industry averages probably aren’t going to do you much good, unless you know the quality of their traffic, the variance and seasonality of their data, and the distribution of the rest of the pages on their site.
Rather, you should look at pages with a similar context within your own website, and then seek to understand “good” and “bad” bounce rates from there. That way, you can ensure you’re looking at numbers within a context you actually understand, and to some extent, can control.
Finally, sometimes a bounce, no matter how you define it, is ambiguous. It’s not clear whether a bounce is good or bad all the time.
For instance, it’s common to gauge the effectiveness of knowledge base articles (as well as other content pages) by engagement metrics like bounce rate and time on site.
But why does a user bounce from a knowledge base article — because they’ve found their answer and are ready to get back to their life; because they tried to follow your instructions but were confused so they left, frustrated, to look at competitors; or because they simply fell asleep while working from home and let their session timeout?
Hard to tell.
For that reason, Optimizely looks both at engagement metrics as well as success metrics, using attitudinal indicators and customer feedback, to see if their educational content is actually succeeding at its purpose.
Just keep in mind that, while a bounce is recorded the same no matter what by your analytics tool, it could be triggered by a variety of user interaction, including:
- Returning to search results, because you didn’t find what you were looking for
- Closing the browser, because you’re cleaning your Mac up
- Entering a new URL in the address bar
- Following an outbound link from your blog post
- Staying inactive and timing out the session
- Reading the whole blog post and then leaving, fully satisfied with the content you read
That’s what I mean when I say that a bounce isn’t inherently good or bad in any circumstance. It really matters on the context (and in reality, a bounce rate is usually a proxy metric anyway. It’s better if you can optimize for metrics closer to business results).
PPC Bounce Rates and How to Analyze Traffic Sources
How you treat bounce rates should also depend on what your traffic source is. If a blog post gets a lot of organic traffic but has a bounce rate, that’s a different story than running Google AdWords and having a high bounce rate. Similarly, if you go viral on Hacker News and have a high bounce rate, it’s a lot different than if you’re targeting a niche audience via Facebook ads.
If you run a linear regression on bounce rates and conversion rates, there’s pretty much no correlation. Even if there were, you have to keep things in context and think about user intent at different stages.
For example, you can see in the image above that display has a high bounce rate, but about the same conversion rate as other channels. A high bounce rate here isn’t necessarily a bad thing, depending on how you’re spending your ad dollars.
Often times, a display interaction will be user’s first touch point–and they may bounce in the immediate experience, but come back again to convert.
That’s why it’s recommended, if you’re spending money on online marketing and operating in various acquisition channels, to do some basic attribution modeling and at least analyze your conversion paths.
Finally, most paid acquisition marketers recommend you control your landing page traffic as much as possible. If you can, it would be ideal to use dedicated landing pages and not index them in search engines. In that way, you can exert more control over click intent and the temperature of your offer. Meaning, you know all traffic coming in for that landing page is paid, and not a mix of paid and organic.
Often, this isn’t possible. For instance, in the case of ecommerce, you’re usually not going to have separate paid acquisition product pages, so you’re going to get a mix of traffic sources. What’s that mean? Basically, make sure you’re keeping traffic source and campaign in mind when you’re analyzing landing pages to research those with low bounce rates.
For instance, if you go to Behavior >Site Content > Landing Pages, and view your report using the comparison feature, you’ll see something like this:
Now, you’re looking for pages with a solid amount of traffic yet have a high bounce rate. Google Analytics tells us here that number three on the list is a good candidate to look into (33% higher than average bounce rate).
Before you jump right in and start changing your landing page up, I’d go into Google Analytics, click the URL that has the high bounce rate, and add a secondary dimension: Default Channel Grouping or Source/Medium. Then, view bounce rate by channel and see what’s up.
In the case of the client above, the high bounce rate on that page was due to the page receiving almost all organic traffic, and the others being heavily paid traffic.
Another point with PPC and bounce rates is that you can use bounces, exits, and shopping behavior segments as remarketing rules. So, if you’re running a paid campaign to a landing page with multiple steps, you can target those who bounce right away differently than those who exit on page 2 or those that complete the conversion (as well as those who trigger any event in between).
When you reframe bounces like this, they become a valuable piece of information you can use to inform your marketing and better align to the buyer’s journey.
How to Lower Bounce Rate in Google Analytics
There are two broad ways to lower bounce rate in Google Analytics: change your definition of bounce rate, or change user behavior on your site.
We talked about the first way already. It includes setting up Google Analytics event tracking in order to set up an “adjusted bounce rate.”
This is a strategic decision with many implications for your future analysis efforts, so it’s important to step back and map out important and meaningful user interactions. Think about what critical things users do on your site that signal engagement (e.g. playing a video or downloading a file). Then, move on to create Interaction events to adjust your bounce rate.
For example, Austin Primary Care triggers an event when you click “Book Online,” and an interactive form pops up:
The second way has more to do with user experience and leading users to interact with your website in different way.
First, let me say that there are ways to sort of “cheat” at lowering bounce rate. By that, I mean it’s quite possible and pretty common to lower bounce rate, but not move the needle on metrics that actually matter like purchase conversions (or maybe even lower that metric).
For instance, say you have a landing page with a form above the fold and a simple submit option. Something like this:
Forgive the poor design, but imagine for a second that this website, instead, removed the “contact us” bar. Now, imagine the only option on the homepage is to click to the next page to find the form submit option. You’re providing a similar conversion path (with one more required step added)–so you may very well reduce bounce rate, but form submissions could stay the same.
An example could be this primary care provider (with much better design):
The same goes for chopping up single forms into multiple steps, especially if the first step only asks for something super easy like “first name.” Yes, more people may fill that out–but it’s possible a similar number will drop off once they see they have to enter their phone number, whether that ask is on step 1 or step 3 of the form (of course, all of this is worth testing).
This is all hypothetical, but what I’m saying is that lowering bounce rate, by itself, is often the wrong way to look at optimization. When you can, look to optimize business metrics.
But perhaps you’re using bounce rate as a proxy to determine high impact optimization opportunities. In that case, you can do some analysis to find pages that are 1) high traffic, 2) very similar in context and format, and 3) have an engagement or conversion focus on page that you can measure.
I previously gave the example on a CXL blog post of clustering landing pages by topic. In this case, I opened a landing pages report (Behavior > Site Content > Landing Pages), and then filtered by URL. Then, I used the “comparison” feature to look at bounce rates compared to the site average. There, it becomes clear which landing pages are underperforming, and the benefit here is that the context and traffic acquisition for all the pages is very similar:
Once you determine landing pages that have lower than average bounce rates, you can then begin to research why that may be.
Maybe the traffic you’re bringing to the page with paid Facebook ads is expecting something completely different than the page offers. Maybe there’s a bug that prevents Internet Explorer users (is anyone even still using I.E.?) from converting on one of the pages. People often forget there’s more than one facet to landing page optimization — and you may need to tweak your targeting, ad messaging, message match, or the landing page user experience. The possibilities are endless, but it’s up to you to dig in and try to solve those problems.
You’re now embarking on a journey of conversion research, which should be followed by experimentation, analysis, and then iteration (and continuous and never ending improvement, of course).
A Bounce Rate Google Analytics Summary
Bounce rate in Google Analytics is a widely misunderstood metric–but when placed in the right context, it can be useful for analyzing user behavior and website engagement.
At the opposite end of the spectrum, marketers tend to overvalue and misinterpret the metric. Don’t even get me started with trying to measure yourself to industry bounce rate benchmarks.
But if you keep in mind the context of your user intent, off page experience, on page user experience, and the distribution of the rest of your website pages’ bounce rates, it can certainly be a helpful metric for finding prime conversion optimization opportunities.