Simplifying Analysis, Without Dumbing it Down

If I was a promoter of web analytics grudge matches, I'd pit Eric "Web Analytics is Hard" Peterson against Avinash "It's Easy With GA and My Blog" Kaushik.  The bout would prove, to all seven of those in attendance, once and for all if web analytics is easy or hard.  To referee the proceedings I'd send out Gary Angel -- not to scrub the blood, blog comments and twitter feeds, but to shed some perspective.

At Semphonic we've wrestled with the notion of whether web analytics is easy or hard for the better part of a decade and a half (myself for the half portion) and, like a good referee, we come down squarely in the middle -- sort of.  A deep dive analysis is by nature challenging.  It involves pulling data, reacting to the most interesting components of the data, adjusting your data pull accordingly and repeating.  Over and over and over, until you have merged off the superhighway of data down, veered sharply on a surface street and made your way down a single file path to actionable insight.  It is the reaction, the human element, which will always make doing analysis hard.

Semphonic finds itself in the middle ground between easy and hard, not because we'd ever argue that doing an analysis is easy, but because we expect (and demand) that comprehending the results of an analysis should be.  Whereas the inner workings of an analysis might look like gibberish sitting in a folder on the analysts desktop, massive data pulls, an inordinate number of cryptically named segments, notes scattered throughout Excel worksheets, the final result should be crisp, convincing and in VPese.

This translation and transformation from complex deep dive data to straightforward analysis deck is, admittedly, easier said than done.  It's something Semphonic over the years has tweaked, changed entirely, gotten wrong, gotten right, gotten wrong again, but through it all improved upon.  I wouldn't call our approach today perfect (to paraphrase French Laundry Chef Thomas Keller -- perfection is fleeting, once you obtain it, it is already gone), but I will attest that it has gotten pretty darn good. 

The centerpieces of Semphonic's approach to simplifying analysis are two "foundational" techniques:
•    Functionalism
•    Use Case

Functionalism is used to explain the complex inner workings of a website in plain English by chunking a website up based on the purpose, or function, of each chunk.  Essentially Functionalism asks the question 'why did you build X?' and based on the answer classifies the page into a group.  To illustrate this, let's do a bit of role playing --

Functionalism is used to explain the complex inner workings of a website in plain English by chunking a website up based on the purpose, or function, of each chunk.  Essentially Functionalism asks the question 'why did you build X?' and based on the answer classifies the page into a group.  To illustrate this, let's do a bit of role playing --
    Jesse: Why did you build your home page?
    You: To route people deeper into my site.
    Jesse:  Cool, we'll classify your home page as a 'router' and measure it's performance against other     pages we consider routers!    

Meanwhile, we utilize Use Cases to bring in the human element, which is too often missed, into an analysis.  By talking about real groups of people on your web site their size, issues and successes become tangible.  It is one thing to talk about an abstract population segment defined by a user who enters on a Google paid search ad, lands on the home page and immediately navigate to the store locator tool and another thing to discuss the real people coming to your site with the sole purpose of locating a retail outlet.  Good Use Cases don't need to be explained or clarified when presenting them to execs, which makes it all that much easier to say there are exactly X thousand of a type of person, with Y thousand experiencing a specific issue and that the issue is having Z impact on your bottom line.

Functional and Use Case analysis can conceivably be use standalone, baseline analysis, but for me they are best suited to strengthen analyses with more specific focuses (internal search, conversion funnel, etc.) by making them easy to understand.  When used correctly than can become tools that bridge the gap between an in the weeds analyst and upper management.  Some of the best analyses Semphonic have done don't mention 'web' or 'analytics', 'visits' or 'visitors', 'engagement' or 'success metric'.  They leverage Functionalism Use Case analysis and become too business focused to get caught up in irreverent web analytics specific vocabulary.   

If those last lines sound a bit like a sales pitch, it's only because they are.  On July 29th Gary Angel will be hosting a webinar to walk through Semphonic's Advanced Analysis program, where Functionalism and Use Case are core pieces.  The program is designed to make deep dive analysis a central and valuable piece of the decision making process.  Regardless of if you are a manager who is considering Semphonic to guide their advanced deep dive analytics program or a do it yourself analyst, the webinar should be of interest.  As Gary put's it "...be forewarned – this is a sales presentation. On the other hand, it’s not without plenty of real content – perhaps more content than some people’s content presentations." 

Meanwhile, I'm just getting started with Functionalism, Use Case and how us as web analysts can leverage the techniques to keep our final product simple while our processes become increasingly advanced and complex.  I realize this blog was more about telling than showing but, in a series of upcoming blogs, I plan to show how these techniques were used in recent SEO, internal search and survey analysis.  Stay tuned.

Comments

[medieval clothing]

interesting which makes it all that much easier to say there are exactly X thousand of a type of person, with Y thousand experiencing a specific issue and that the issue is having Z impact on your bottom line

[lasersman]

green laser pointer is very cool and convenient, I have bought 3 pens and recommended laser pointer to my friends.

[VectraLasers]

I have to try this tool. Some say this is tool is great.

[Laser Pointer]

Nice tools!

[שיווק באינטרנט]

When using Ood should not be explained or clarified when they submit them to the execs, which makes it all much easier to say that there are exactly X miles for a type of person with a thousand knows Y a specific question and that question is Z having an impact on your bottom line.