Interesting, I hadn't understand that possible meaning in the sentence at all until reading the comments. I had assumed it was a 30 second reduction in page load time or some such (Though I suppose it should've been obvious this wasn't the case... A website with a 30+ second load time would likely not be able to gain a large audience). Either way, a confusing headline...
I suppose this chain of comments just goes to show how differently we all think. That was the first title I thought of to describe the topic of the post in limited space, nothing more.
If you look at the other posts on my site, it should be obvious that I'm not good at (purposely) writing link bait type titles. Sorry if you feel like I tricked you. Wasn't my intention.
I only submitted it here at all because I found Clicky's new bounce measure profoundly helpful. That data would have been invaluable to me years ago. I thought anyone working on a web startup would surely also be interested.
I felt slightly conned, as you say "stunningly good link bait" but not deeply disappointed, the article was interesting, well written and clued me in on a web analytics provider I didn't know of.
This article is about how he redefined "bounce" to not include people that viewed the site for 30 seconds or more.
While I appreciate the sentiment, I consider "bounces" to be a lack of up-sell. My stats give me a "bounce" rate as well as the average time spent on that page for new visitors, so I can already see both of the stats he's referring to. They are just useful for different things.
Not in all cases. For an affiliate site, a bounce could be some one who gave up reading, or someone who was completely sold and clicked through on the affiliate link. They both show up as "bounces" in Google Analytics.
Since I'm not selling anything, a single pageview can easily be a success case for me. If Google refers someone to my site and a single page (and/or its comments) answers their question, I see that as a win. That's where most stats packages fail me, since they assume single pageview visits are immediate bounces.
I think it's applicable to other situations as well. Even if you're selling something and hope to guide visitors down a funnel, an accurate time-on-last-page metric should be valuable. There's a world of difference between someone that bounced immediately and someone that read your landing page for a few minutes but wasn't convinced to take the next step.
If you just want single page view stats, my stats give me "time spent on site" broken down by "new visitors".
I don't understand why you have to redefine the term "bounce" in order to achieve that. You could simply say that "bounce" wasn't the metric that you were looking for, or that people put too much emphasis in it. You don't need to "fix" it because it isn't broken.
"Bounce" is a measurement of people that do not continue to interact with a site after the first page they visit.
What stats package do you use that gives you a "time spent on site" that's accurate for a single or final pageview (honest question)?
Most of them take a very pessimistic view. I believe Google Analytics counts a single pageview visitor as 0s, for example. That turns out to be terribly inaccurate for my site.
The problem is that "bounce" is a broken metric. It implies that people "bounced away" from your site very quickly because they didn't find it engaging. But the standard definition of a bounce is simply a single pageview visit. It needs redefining, because there are plenty of single pageview visits that do actually engage the reader. e.g. reading this article - I bet a lot of you read it but didn't read any other articles on the site. GA would probably report 90% bounce rate for these visits, but Clicky would be more like 20-30%, which is much more accurate measurement of how many people are actually engaging with the site.
Obviously, the key question here is what the desired action on your landing page is.
If you're Hanselman, and your goal is to get people to read your blog posts, then your bounce-rate can be close to 100% and it doesn't really matter-- the content you are attempting to serve up is all on the front page, and additional clicks are secondary.
If, on the other hand, you have an e-commerce site like Amazon's, where the landing page is just the first step in a process that you'd like to end in a conversion, then a 78% bounce rate is definitely something to work on.
While the original 78% bounce rate missed people who happily read the full page and then were done, the new 15% "<30 seconds on page bounce rate" overcounts people who leave the page up without reading.
When discovering new material (as from HN or long lists of search results), I'm sure I'm not alone in that I open a bunch of tabs, then slowly work through them. Sometimes, when I reach one, even if it's been open for minutes (or hours!), it's dismissed immediately as uninteresting or obsolete.
I've noticed some video sites manage to defer auto-start of the video until I bring the tab to the fore. Other sites seem to only load (or fade-in) inline images as I scroll to them. Has any analytics tool added similar monitoring of foreground-time or scroll-to-end-of-item sensing?
I believe Clicky only continues pinging for 10 or 20 minutes after a page loads (not sure which; they've changed it up a bit since introducing the pinging).
That's a great idea about detecting whether or not the tab is actually active. I don't know how difficult that is to implement without a visible element on the page, but I'm going to ask Clicky about adding that feature.
I agree about the overcount vs undercount thing, but I think Clicky's method is much closer to the "truth" of site engagement. Tabs aren't used much outside of the "power user" world. Typically when people go to a page, it is the page they are reading.
We do want to make it smarter and only ping when the tab is in the foreground, as that will make it more accurate. It's a planned update for the future.
I really wonder how useful the time-on-page metric is. Thanks to tabbed browsing, I might have your site open for 30 minutes, yet only look at it for 30 seconds.
In Google Analytics, under Visitors > Visitor Loyalty > Length of Visit, you can find the number of 0-10 second page views, which could be a useful metric to try to reduce - now if I could make a line graph out if it instead of a bar graph for the period. Anyone know if this is possible?
28 comments
[ 3.9 ms ] story [ 52.9 ms ] threadIt's really hard to come up with any other interpretations of the subject line a priori.
It's stunningly good link bait, and the article is deeply disappointing compared to the promise.
If you look at the other posts on my site, it should be obvious that I'm not good at (purposely) writing link bait type titles. Sorry if you feel like I tricked you. Wasn't my intention.
I only submitted it here at all because I found Clicky's new bounce measure profoundly helpful. That data would have been invaluable to me years ago. I thought anyone working on a web startup would surely also be interested.
While I appreciate the sentiment, I consider "bounces" to be a lack of up-sell. My stats give me a "bounce" rate as well as the average time spent on that page for new visitors, so I can already see both of the stats he's referring to. They are just useful for different things.
Not in all cases. For an affiliate site, a bounce could be some one who gave up reading, or someone who was completely sold and clicked through on the affiliate link. They both show up as "bounces" in Google Analytics.
I think it's applicable to other situations as well. Even if you're selling something and hope to guide visitors down a funnel, an accurate time-on-last-page metric should be valuable. There's a world of difference between someone that bounced immediately and someone that read your landing page for a few minutes but wasn't convinced to take the next step.
I don't understand why you have to redefine the term "bounce" in order to achieve that. You could simply say that "bounce" wasn't the metric that you were looking for, or that people put too much emphasis in it. You don't need to "fix" it because it isn't broken.
"Bounce" is a measurement of people that do not continue to interact with a site after the first page they visit.
Most of them take a very pessimistic view. I believe Google Analytics counts a single pageview visitor as 0s, for example. That turns out to be terribly inaccurate for my site.
If you're Hanselman, and your goal is to get people to read your blog posts, then your bounce-rate can be close to 100% and it doesn't really matter-- the content you are attempting to serve up is all on the front page, and additional clicks are secondary.
If, on the other hand, you have an e-commerce site like Amazon's, where the landing page is just the first step in a process that you'd like to end in a conversion, then a 78% bounce rate is definitely something to work on.
When discovering new material (as from HN or long lists of search results), I'm sure I'm not alone in that I open a bunch of tabs, then slowly work through them. Sometimes, when I reach one, even if it's been open for minutes (or hours!), it's dismissed immediately as uninteresting or obsolete.
I've noticed some video sites manage to defer auto-start of the video until I bring the tab to the fore. Other sites seem to only load (or fade-in) inline images as I scroll to them. Has any analytics tool added similar monitoring of foreground-time or scroll-to-end-of-item sensing?
That's a great idea about detecting whether or not the tab is actually active. I don't know how difficult that is to implement without a visible element on the page, but I'm going to ask Clicky about adding that feature.
We do want to make it smarter and only ping when the tab is in the foreground, as that will make it more accurate. It's a planned update for the future.
Also, is it possible for javascript to check whether the site's tab and browser window is viewable and in the foreground?
I want to know what should be accurate bounce rate of a blog ?? marck_don
http://www.adnpost.com