I had the pleasure to see "Twelve Angry Men" at the Ahamson Theatre this weekend with George Wendt and Richard Thomas - what a wonderful play to see.

But it got me to thinking about Gary Angel’s recent post entitled "Web Analytics Toolkit and Internal Search". Commenting on a Commerce360 blog entry that laments the lack of presentation on how to do the analysis part of web analytics, Gary says,

"So true. And the comment doesn’t just apply to Conferences. It applies to the vast majority of all our blogging as well. In fact, I find that even when most of us are talking how-to analytics that the discussion might be about some charting technique, data visualization or data manipulation. That’s all well and good, but it takes for granted that you know how to actually use that technique to approach an analysis."

Why do these two things go together in my head (besides the random association that happens more than I’d care to admit)?

Well, in the play (spoiler alert for anyone who hasn’t read or seen it), a jury is presented with a set of facts that they must use to determine the guilt or non-guilt of the accused. Each of the panelists, for reasons of their own has arrived at a decision - at least they think so. Eleven them, for a variety of reasons including the most trivial (baseball tickets) to reasoned logic, have decided, at the beginning of the play that the accused is guilty. But there is a lone standout - who is preventing the other eleven from returning to their ‘normal’ lives.

This dissenter (played by Thomas) proceeds to take the other members of the jury through the case, point-by-point, and evidence-by-evidence and questions each one. He pokes holes in everything that was presented by the prosecution and even tears down the public defender. By the end, he has convinced the other eleven that there exists reasonable doubt that the defendant is guilty and it’s twelve votes for not-guilty.

What did he do and what does this have to do with the analysis part of web analytics?

He observed, and prodded, and poked each datum, he ran scenarios in his mind, he found inconsistencies in the logic (as presented during the trial) and the facts. In a word, he analyzed.

It’s interesting to note that eleven jury members all started out with the same decision (guilty) but for a variety of reasons - they didn’t all get to that decision in the same way.

So, why don’t we blog about or teach analysis? Because it’s hard (where have I heard that before). My way of doing analysis is probably different from just about everybody else’s.

My basic approach is to look for patterns and violations of pattern. When I find a pattern or a violation of a preexisting pattern, I have something tangible, something that I can try to explain. My patterns are built out of a lot of things: my familiarity with the data, my familiarity with the website, my familiarity with the market, my familiarity with the data collection tools, my accumulated experience and on and on.

It’s also a function of my training, I studied as an anthropologist in school and my early career was spent doing technical support. Technical Support is essentially analysis of a system in order to identify a root cause and eliminate that cause so that the system can work correctly. That training BTW, provides my primary framework for analysis (at least I think so).

I don’t think the above is particularly teachable - you can read the words above, and believe that’s how I do it, but how much does that help you in learning to do analysis?

One last thought…

Let’s also take a look at the definition of analysis.

According to Dictionary.com, analysis is "the separating of any material or abstract entity into its constituent elements (opposed to synthesis)." Isn’t that interesting? There are of course many more definitions, but this is the first one. It seems to me that analysis in this context is breaking down behavior into it’s constituent pieces - hmmm isn’t that we do in the data collection phase?

Here’s the definition of synthesis: "the combining of the constituent elements of separate material or abstract entities into a single or unified entity (opposed to analysis)." Ok, so they language is fairly opaque, but what it says to me is taking a bunch of discrete elements and putting them together in an understandable and informative way.

…I think I’ll go ask my boss if I can be a web synthesist instead of analyst