Judah Phillips’ recent Mediapost post Web Analytics: Easy? Hard? Complex? It Depends is an excellent, well-rounded discussion.
Judah discusses in depth how and why analytics being easy, hard or complex depends on
“… the tools you are using,
the site you are analyzing,
your company’s requirements,
your team’s skill set, and
the processes you define for analytics.”
I agree with Judah. It does depend. And there is something you should do to make Web Analytics easier:
Ask good questions.
- Good questions are those whose answers matter.
- Good questions are those whose answers lead to action that matters.
- Good questions are those whose answer lead to action that results in measurable business impact.
Pulling data without a question in mind often leads to frustration. There will always be differences in the data. But if the differences don’t matter, nobody will take action and you’ll be frustrated by the time you’ve spent pulling and segmenting data.
So find out what matters first.
I was just about to post, and in the time I’ve been writing this response to Judah’s post, a comment has been posted by Eric Melchor from Smart Digital Spending that speaks to this frustration.
Eric asks “But after hours, days or even weeks pulling reports, how is knowing that visitors stemming from organic search stay 30 seconds longer on the site versus paid search visitors going to change the way you buy media or site design? How is knowing that visitors from Fargo tend to have a higher bounce rate going to change the way you design your website?”
To avoid frustration, first find out what matters to the client in the way of results, and what type of action they expect to take. Ask questions before you begin pulling data and analyzing. Ask questions so that you have a useful analytical context of what matters.
Using Eric’s questions above, hypothetically speaking:
- If the client is interested in shifting media spend, rather than starting by analyzing the metric ‘time on site’, a question about how media effectiveness is judged would likely have led to a comparison on value events (whatever they are) between media rather than a comparison using the ‘time on site’ metric.
- Regarding visitors from Fargo …. Are visitors from Fargo part of the target audience?If we knew where we expected our most valuable visitors to be coming from and Fargo was at the top of the list, a high bounce rate should indeed be a concern.If Fargo is not at the top of the list, we shouldn’t even look at the bounce rate for Fargo because there are more important visitors to focus on (and if visitors from Fargo are not the target audience, a high bounce rate may confirm web site design is right on.)
So, it depends.
What do you think?