No single metric exists in a vacuum. You should always look at one metric in the context of other metrics. Once you start measuring in terms of rates, the depth of context is multiplicative. Your talking about two metrics each with their own contexts. An example of the context around one metric is visits. Visits may be up, but you must remember that visits are collections of page views. It might be a great feeling to report that visits are up, but what if page views are down? For some that might still bode well, so you have to take into account the intent of your website.
“Come on, Dustin, what is the missing factor of bounce rate?” Ok, ok…it’s the total number of visits to that particular page. Without that context, you run the risk of providing an empty insight.
YOU: ”The bounce rate on the FAQ page is 90%. We better fix that!”
SOME OTHER PERSON WHO ALWAYS PICKS AT YOUR ANALYSIS: ”Hey, our FAQ page only gets ten visits a month! You should have had a V-8! [sound of your head being smacked]”
So don’t forget context and breakfast when considering bounce rate and other metrics.
I remember a while back a huge debate on how to measure engagement. I remember Eric Peterson’s huge formula that reminded me of mathematical analysis courses I had recently survived. Someone (I seem to recall it being Avinash Kaushik) was very much against this. And so the debate began. In my opinion, Avinash (or whoever it was) made the most sense. Boiling it down to one number is not very helpful. Directional change is meaningless with so many variables involved. Proving this would amount to some mathematical analysis of limits I’m sure.
Taking this historical anecdote into account, can you imagine my response to TweetLevel‘s importance metric? What makes your twitter activity valuable is different from what makes mine valuable. Suppose I’m using mine to promote my blog and you’re using yours to just tell people what’s on your mind. I would ultimately be interested in clicks back to my blog, while you may ultimately be interested in how many followers you have (while using the number @reply’s as an indicator of engagement).
Any time I see anything more complicated than a rate, I cringe. Don’t hide meaning. What if you had an engine similar to Google Analytics new Intelligence that could tell you what has changed and why it is important? Remember this: What you measure is determined by why you are doing what you’re doing.
