I owe the web analytics community an apology.  I was doing some analysis yesterday that made me realize that my Omniture formulas for the hard bounce rate and the soft bounce rate didn't match my definitions.  So here are my corrected formulas:

• Soft Bounce Rate:  ( [exits] - [single access] ) / ( [visits] - [single access] )
  
• Hard Bounce Rate:  [single access] / [entries]

Originally, I had the soft bounce rate as ( [exites] - [single accesss] ) / [visits], then I realized that the definition requires the denominator to be only the multi-page visits.  In light of this, the formula can now be stated as the number of exits that were not part of single page visits divided by the number of visits that were not single page visits.  Here we are trying to answer the question, "Of visitors who have visited at least one other page before this one, what percentage end their visit on this page?"

I also incorrectly had visits in the denominator for the hard bounce rate, when it should be entries.  So the formula in plain english is calculating the number of single page visits divided by the number of entries through that page.  Here we are trying to answer the question, "Of visitors who start on this page, what percentage don't visit any other pages?".

One caveat I have thought of is when a visitor visits a page, refreshes the page, then leaves.  It is advised that you consider the refresh rate when contemplating the soft bounce rate as it will add to the soft bounce rate when you would probably consider that a hard bounce.  Although, not every refresh is a two page visit.

I was on a call yesterday with a web analyst from our implementation team looking into savvy ways of segmenting web analytics data by survey responses.  Lo and behold, the survey page wasn't tagged with our web analytics code!  Guess what success event on the page resulted in an increased exit rate?

In short, here's a best practice when analyzing and reporting exit rates.  Answer the questions, "Does the page link to pages that should be tagged but aren't?"

I've been hearing a lot of talk lately of bounce rates, exit rates, puke factors, bail-outs, etc.  If you're confused like I was trying to figure them all out, then read on...

The first thing that will help is to understand how time spent on page is calculated in most web analytics tools. When a page is visited, a time-stamp is written in the browser's cookie to identify when the session began. When the next page is visited, another time-stamp is written in the browser's cookie. If a visitor visits one page on your site and leaves, most web analytics tools will not know how long they spent on that page.

Understanding time spent on page is crucial because one way of measuring the bounce rate is by the percentage of "short" page views. "Short" being defined by page views less than a set time, such as five seconds.  Measuring bounce rate like this is not relevant in most cases.  We're limited to counting pages in visits rather than keeping time.

Exit Rate:  The percentage of visits including the selected page in which this was the last page visited.

Soft Bounce Rate:  The percentage of multiple page visits in which the selected page was the last page visited.

Hard Bounce Rate:  The percentage of visits including the selected page in which this was the only page visited.

Note:  Exit Rate = Soft Bounce Rate + Hard Bounce Rate

For Omniture users:

Surprisingly, these are not readily available in Omniture and require some tricks to find.  Fortunately, there is a feature called the "calculated metric".  Following are instruction on how to set these up as calculated metrics in Omniture:

Note before you begin:  You will need to set these up in each report suite that you use.  :(

• Open the Traffic > Segmentation > Most Popular Pages report.
• Click on the metric that appears underlined and in bold after the word "showing" on the upper-right corner.  This will open a window in which other metrics may be selected.
• Click on the "calculated" tab.
• Click the "Define New Metric" button.
• In the new window, name your metric accordingly.
• Select "Percentage" in the upper-right corner.
• You may also define the number of decimal places.
  
• You will be using the "standard metrics" in the left column to build your calculated metric.  Here are the calculations for the metrics presented:
• Exit Rate:  [exits] / [visits]
• Soft Bounce Rate:  ( [exits] - [single access] ) / ( [visits] - [single access] )
  
• Hard Bounce Rate:  [single access] / [entries]

The larger the company you're working with, the more likely that you will see this occur.  A conversion path may consist of pages that are owned by different organizations within the company.

For example, the product support page may be owned by the product's web content team, while the more detailed support page is owned by the software groups support web content team, and yet the purchase pages are owned by the company's online purchase web team.  Yikes!

The good news is that all three teams are tied to the success of the entire conversion path even though they only have ownership of a segment of it.  My suggestion is for all three teams to come together to look at the conversion path heuristically and each plan to make one significant change, then make the change, measure the impact and regroup.  This process should be repeated until all parties are satisfied with the performance of the entire conversion path.  Then, it is important to continue measuring and return to this process if necessary.

The other option is for your team to improve your portion, then ask for the other teams to participate.  Definitely improve your portion of the path before you go knocking another teams portion!

Today I was asked a very nice question and figured I'd post my response here:

Q:  "What, in general, do you think is an acceptable bounce rate for an index page that gets a lot of entries?"

A:  Of course the lower the better in this case.  The key is to take what you have as a baseline and try to improve it.  Most importantly, try one thing at a time.  It will take some brainstorming to try and figure out what might be confusing to visitors or what can be cleaned up.  You can also use ClickMap and Google Analytic's site overlay feature to see what is going on.  Here are some questions off the top of my head that might help:

• How do visitors arrive at this page?  If there are visitors arriving by mistake, weeding them out automatically improves your bounce rate.
• What are the most popular sections of the page and are they most visible?
  

It really involves two pieces - getting pre-qualified visitors to your page and doing a good job at helping them find what they're looking for on the page.  You don't have to do an all-out usability study for changes you'd like to make.  Find random people in the hall and ask them to perform a typical task on the page and get their feedback on which way was simplest.

That's my brain dump.  ;)