≑

A Major Flaw in Google Analytics: Bounce Rates and Total Visit Times Explained

Wifi

bounce rate

Ok, ok, you caught me. I’m using a link bait title to get your attention to get you to read my article. What I’m about to tell you isn’t a flaw in Google Analytics so much as it’s a flaw in how we think about it (although they could certainly do a better job explaining it). This is one of the single most misunderstood concepts I can think of when people view their site metrics. In fact, up until about a week ago, I was one of those people who misunderstood this concept. Thankfully, I am now enlightened, and it is time to pass the thoughts on to you.

So, raise your hand if you have a website and your bounce rate is above 50%. How about 60%? 75%? 90%.

Sounds pretty damn awful, doesn’t it, having a bounce rate above 90%. Now raise your hand if you can actually tell me what bounce rate means without going a quick Wikipedia search? Hint: it has nothing to do with how many seconds someone spends on your site. We’ll get back to this in a second.

As long as we’re on the topic of awful things, who here can say that they have visitors who, according to Google Analytics, spend 0.00 seconds on your website? Everyone’s hands should be raises at this point, cause everyone has those weird non-visits, at least according to Google.

What’s going on here?

The problem that we’re experiencing is simply in the incorrect way many of us assume Google Analytics is reading our webpage. Here’s an example to illustrate how it really goes:

When a user enters your site, a page load request is processed by your server. This request is noted by Google Analytics and is officially logged as the start browsing time. Let’s pretend that a user loads your page at 12:00 PM. This is noted in the log.

At 12:05 PM, the user finishes reading your homepage and decides to click on a tab on your site and load a new onsite page. This page load request is noted by Google Analytics and the time is recorded. At this point, and at this point only, Google Analytics can now report that the user has spent 5 minutes on your site.

At 12:08 PM, the user clicks an internal link in your site and begins reading a new page on your site. Google Analytics logs the page load request and now know that the total time on site is 8 minutes.

Now pretend the user spends 30 minutes on this third page, carefully reading a long article you have posted. When they finish your article, they click off your site, go to a new website, or click on your affiliate link. They don’t load another page from your server. What just happened..? There is nothing in the log file to indicate that they loaded another page from your site, so Google Analytics doesn’t know that they just spent 30 minutes on your last page! In your Google Analytics report, you would see that this user spent 8 minutes on your site and viewed two pages. As far as Google is concerned, that last 30 minutes never happened. Your server doesn’t make an announcement that a page has been ‘unloaded’.

If you haven’t seen the obvious problem in this, I’ll spell it out for you: What if your website just has one landing page??? The user will never have a reason to click away from your one page. They’ll arrive at your website, view your page, and either click on a link or hit ‘Back’ in their browser. As long as they don’t load another page on your website, there is no log of how long they spent on your website. This visitor has, according to Google Analytics, just spent 0.00 seconds on your website. They are also CONSIDERED A BOUNCE. A bounce, by definition, is a user that only loads one page of your website! This might be useful for people with e-commerce websites, but for those of us with information or content-based sites, what happens when the user lands on the page, finds the information they are looking for, and immediately leaves? They are contributing to our bounce rate with no real indication of how long they were on the page and whether or not they found what they are looking for. Also, the total time on site of ANY visitor will also be under-reported by one page’s worth, (the last page loaded) and whether they spent 30 seconds or 30 minutes on that page be never be known.

Now, you can see how many campaigns to ‘lower your bounce rate’ or ‘increase time spent on site’ fail. Many people are operating under the assumption that bounce rate refers to people who leave your site after just 10 seconds, or just 30 seconds. They spend hours putting pretty images, enticing displays and other imaginative ways to keep users on their site. Unless those fancy methods encourage the user to click through to other pages on your site, however, they aren’t going to really have an affect on your bounce rate and you’ll be unable to tweak conversion rates for a home page that has, say, a lead form.

There are various work-arounds for this issue, if you are inclined to try, but none of them work very well. Many of us, for both personal ego reasons and business-related reasons, would like to know how much time users are ACTUALLY spending on our website, but it’s just not feasible at this point in time for Google Analytics. Some people have tried working with page unload tracking with moderate success, and you can find a little detail of their efforts here.

This makes me have major issues with Google Analytics and question its feasible usage. If I am to continue using it, I would definitely consider eliminating Time on Site and Bounce Rate as relevant metrics of performance evaluation. If clients still want to know about their statistics on these metrics, I would have to explain to them why they aren’t very good ways to evaluate performance.

The original article that inspired me to write this piece can be found here.

Image Credit: ddablogimages.

Article by David Fishman

David has written 3 articles.

You can follow webtrafficroi on Google Plus, Facebook and Twitter here. Subscribe to WebTrafficROI feed via RSS or EMAIL to receive instant updates.


Comments on this entry are closed.

  • Mathew Day July 23, 2010, 4:18 pm

    This is one area I didn’t know much about. But after reading your thorough article, I now understand bounce rate and how Google Analytics tracking works.

  • Uzi July 23, 2010, 8:11 pm

    Some companies define it as, “single page views”. I hope Google can fix this problem in future by using AJAX.

  • SEO: Top Commenter July 23, 2010, 9:25 pm

    I know that a lot of people are suffering from the same the same thing you did. I think I got lucky. The first time I ever looked at GA, I saw bounce and then look it up. I didn’t get sucked in – whew. You should know that you can attach actions to GA on click events though. This allows you to track click-out exits on a landing page.

    • David July 26, 2010, 7:17 am

      Yes, you can track exit links, but most people are just going to hit ‘back’ in their browser or load a new page URL. πŸ™

      I suppose you could track exit links and subtract that from your bounce rate to get your ‘true’ bounce rate. StatCounter tracks exit links. I have both GA and SC running on a few of my web properties.

  • ZK July 24, 2010, 12:35 am

    I want to thank David for writing the very first article on web analytics for this blog.

    Internet is the only business where you can measure everything about your website , users and business. In the coming weeks we will be focusing more on web analytics and how it can add value to your online business.

  • Giochi July 24, 2010, 7:29 am

    I didn’t know about this.
    I hope it will be fixed soon.

  • Dean Saliba July 24, 2010, 11:42 am

    I actually wrote a post recently saying using fake post titles can hurt your blog and then the first blog I check out has one. πŸ˜€

    • David July 26, 2010, 7:23 am

      @Dean

      It wasn’t technically an entirely false post title, as you railed against in your 3rd point in your post. It was just an..overestimation of the total impact of said topic πŸ™‚

      I certainly hope you got beyond the opening paragraph after you discovered I had ‘tricked’ you πŸ˜€

  • Jay July 24, 2010, 12:00 pm

    My blogs have very high bounce rates – around 75% which I’m not proud of. I was previously aware of page views impacting the bounce rate. One question that I have is that my bounce rate is super high as measured by Google Analytics, but it’s actually much lower when measured by Quantcast? Do you know what the reason is for this discrepancy?

    • ZK July 26, 2010, 12:13 am

      Have nt used Quantcast , but every analytics tool will show you a different reading for visits , bounce rate , page views etc. There will always be a slight variation in the numbers

      • David July 26, 2010, 7:45 am

        I haven’t used Quantcast either, but I’ll first agree with ZK’s statement about normal variation in the numbers.

        Next, I’ll offer a theory. I don’t believe Google Analytics tracks exit links, so there’s no way to differentiate between people that truly ‘bounced’ away from your page and people that clicked on, say, a banner ad or Adsense ad. If your site is made for Adsense, obviously you want your bounce rate to be high. If Quantcast can track which clicks are on your banner or text ads, it would probably subtract that percentage from the overall bounce rate.

  • Plumber Sydney July 25, 2010, 4:40 pm

    Bounce rates are extremely important in maintaining rankings and will only become more important in the future. If you have low bounce rates you are obviously providing something of value to your visitors so Google considers you worthy of holding the position in the SERP’s. Thanks for the write up.

    • David July 26, 2010, 7:49 am

      Yes, we know you want to be on the Top Commentator sidebar.

      • used tires July 26, 2010, 9:13 pm

        @Plumber, this article didn’t really touch on the SEO aspect of Bounce Rates… Nonetheless… I would have to disagree with your comment. I think it’s a myth that Bounce rates are a metric that will keep you at a certain position on Google. I’ve seen a video before of Matt Cutt’s who has addressed this issue, and he said there is no correlation between Google Analytics and the Google results page. He said that Google isn’t spying on your Analytics page and using the numbers from there to guide it’s search engine results.

        -Jean

        • David July 27, 2010, 9:25 am

          @used tired

          I was going to respond in the same way you did, but then I did a little searching around and wasn’t so sure. SEO Black Hat did a study apparently and found some interesting things:

          http://seoblackhat.com/2008/11/21/bounce-rate-seo/

          • used tires July 29, 2010, 2:09 pm

            Interesting stuff David, well theres a few ways I could now take Matt Cutt’s video…. perhaps he was right in that Google doesn’t directly use their Anayltics Data, but perhaps what he failed to mention is that they instead use some other method outside of Analytics?

            Till then,

            Jean

  • used tires July 25, 2010, 10:20 pm

    Yeah I had always wondered why there are 0 second times showing up on my blog, always seemed strange as to how someone could consistently record those kinds of times, I thought for sure something went wrong. I guess the end result answer for me personally is just to ignore those.

    Till then,

    Jean

  • ZK July 26, 2010, 12:10 am

    60 % is an average bounce rate for any page but if its a landing page for a PPC campaign then you need to check the content and other visual elements of the page to get it down

  • Property Marbella July 26, 2010, 12:31 am

    Hi David
    The advantage is that it is the same for all web pages. The disadvantage is that I may never know exactly which pages my visitors read the most.

  • Green Planet July 26, 2010, 1:04 am

    I have very awful bounce rate. Most of my visitors are out from the index page as google Analytics says. I have a bounce rate of above 70%, can you please write a post on decreasing this? I know that google analytics is just an estimate, but I really believe the data they show.

    • David July 26, 2010, 8:00 am

      Well Green Planet, you have to ask yourself about each page of your website. I would suggest you install a second web analytics tool like StatCounter so you can track which pages people exist your website from and if they exit by link.

      Maybe you provide a juicy-looking link that your readers just can’t ignore that causes them to leave your site. Maybe it’s something about a certain page that they get to that they don’t like the look of and causes them to leave. Maybe the titles of your posts aren’t very interesting and don’t entice them to click

      The only way to check is to play around with things. Tweak your website until you change the results on your website.

      One thing i noticed when I went to your website is that your theme is very cluttered. My first impression of the website when I viewed it was that it was very ‘busy’ and disorganized-I didn’t know where to click.

  • Dany Event Planning July 27, 2010, 2:49 am

    I have been installed for a week now and I am finding the information both fascinating and confronting. The most confronting one is that Page views doesn’t mean readers.

  • Social media optimization July 28, 2010, 3:48 am

    bounce rate is important to decrease, if you are checking and monitoring your webiste and checking that your website bounse rate is decreasing that means that you are offering some thing good for the customers. you must have to keep good contents, that they have good ideas and new and unique technique what user wants.
    thanks

  • Jennifer @ vergent t1 service July 28, 2010, 10:44 am

    Oh, WOW. Finally, somebody has coherently explained this! Those 0:00 site time loads drive me crazy. What drives me even more crazy is knowing that Google is now using the time on site metrics and bounce rate to calculate the quality of the page, which of course, affects the site’s ranking. Personally, I think if someone comes to your site and finds what they’re looking for on the first page, that’s a GOOD thing. It’s good for conversion rates. . . but bad for SEO rankings. That really puts commercial site owners in a catch-22, doesn’t it? Not cool.

    • David July 29, 2010, 7:50 am

      Hi Jennifer,

      A few comments up, I posted a link to an article/study done at Black Hat SEO that examined the question of bounce rate affecting rankings. After doing more research on it, I’ve discovered that most people are pretty divided on the issue, with most people who think it DOES affect rankings citing that very same article and most people who DON’T think it affects rankings attacking the methodology and interpretation of that article.

      Either way, it would be something interesting to look into more. Anyone have any ideas on how to test this quantitatively? We just need a way to manipulate bounce rates consistently.

  • Private School Marketing Guy July 28, 2010, 1:35 pm

    While this is interesting and many other respectable sounding people take your comments at face value, what proof do you have that this is true? What do they consider an ‘Exit’ then? You would think that a company as smart as Google would be well aware of the flaw here and deal with that. I know that there are flaws with Google Analytics but your example of the 30 minute page read seems to indicate the potential for very bad Time on Site and Time on Page data. Call me naive (I don’t think I am) but that doesn’t sound true to Google’s character. Why would they give these averages knowing they are wildly inaccurate?

    • David July 29, 2010, 7:41 am

      Well, in my first paragraph, I explained that it isn’t so much a problem of how Google analyzes data, but how we interpret it. If you go hunting through Google’s Webmaster Tools, you’ll find that they explain what bounce rate is and how site time is calculated. The phrases are inherently misleading, however. I have always assumed that bounce rate had something to do with people that went to my page and ‘bounced’ away within a certain time frame (for example, 10 seconds or something).

      I also assumed that time on site would be an entire estimation of time on site. If you think about it, however, there’s no realistic way for Google to track what time a page is unloaded from your server. There’s no realistic way to expect Google to be able to track that with their code, and it has nothing to do with naivete or Google’s ‘character’, it’s just reality.

      The purpose of this post was not to call out Google for doing something wrong (although I wish they would get with the times and show me my exit links!) The real purpose was just to point out a common misconception about what exactly this data means for you and your site when you see it on your Analytics dashboard.

  • Online Reputation Management July 30, 2010, 6:22 am

    if you are checking and monitoring your webiste and checking that your website bounse rate is decreasing that means that you are offering some thing good for the customers. you must have to keep good contents, that they have good ideas and new and unique technique what user wants.

  • Sourish August 1, 2010, 9:56 am

    my sites bounce rate is quite low , below 50 for the last few months since i added YARP related post plugin

  • Everton Dantas September 7, 2010, 9:01 am

    Thank you so much for making this clear!

    I have at least 5 major landing pages on my site and couldn’t understand why the bounce rate on some of them was so high, especially after noticing that a visit I made using a very specific keyword was considered a bounce, even though I stayed on the page for 5 minutes!

    Thanks again for the great article!

  • Susan September 26, 2011, 5:19 pm

    This is great to know – I had no idea! I was just reviewing some stats that dumbfounded me and found this article after suspecting it had to do with GA. Thanks for this – it IS a major flaw.

  • David @ Buy Books February 10, 2012, 4:00 pm

    You can be excused for the title – I’ve seen much worse “sensational but misleading headlines”. πŸ™‚

    I find the avg time on page to be the dominant metric when I review my pages. I place less credence on bounce rate – now that I’ve read your article, its significance is even lower.

    For my sites, I’m just happy to see my successful pages with a decently high avg time on page. I am not so concerned if they came from another page on my site, or from the search engine results, or if a given page has a high exit percentage – as long as I have engaging pages (good avg time on page), I’m happy. I am not happy with my “loser” pages though, and these are targets for me to improve upon.