Bad Metrics - Bounce Rate, Time on site and Google Panda Recovery Metrics

Recovering from Panda is all about improving the metrics that Google uses to measure satisfaction with your site.   So, as you have set forth on your aggressive Panda recovery plan (or Panda inoculation plan if you have not been hit) - you naturally are trying to figure out what to measure.    The metrics that everyone is obessing on is 'Time on Site' and 'Bounce Rate'  as they are reported by  Google analytics.   However, these metrics are extremely flawed and if you are looking at these measures it is important to understand the flaws.
Time on Site - The real time a user spends on your site is an obvious and clear measure of the success of your website.   However there is a fatal flaw with the way Google reports Time on Site.   Google does not include the last page of a visit (since it has no way of telling when that page view ended).  This, of course, excludes bounces as well, - so Google does not include bounces when caclulating average time on site.   
I was talking to somebody on a Pandarzied site that had a average time on site of 4:30 - how could they possibly be hit he wondered?  I explained - that 4:30 only includes your happier users, users who liked what they saw in their first page view and decided to click on something.
Bounce Rates -  Bounce rate is yet another flawed metric.  A bounce rate as measuered by Google Analytics is simply a single page view visit.   Perhaps a user  landed on your site,  did not like what they saw and bounced right back to Google in 15 seconds - or perhaps they stayed 90 seconds reading the great content on your page .  Both those visits would be counted as a bounce.   To make matters worse, on-page JavaScrupt or Ajax interaction would not negate a bounce either since they don't fire another page view.   So again, we have a highly flawed metric that is being obsessed over by SEOs, web devlopers and site owners.

Better Metrics - In my next post, I look at other otpions to the standard way of measuring these 2 critical metrics.