The SEMphonics president describes a rich, formalized web analytic strategy, and how to implement this methodology for maximum responsiveness.

Why does the practice of web analytics seem like it should be so rich yet often turns out so fruitless? Other than high-level traffic analysis, site satisfaction scores and the inevitable "we could be doing better with conversion" reports, web analytics are often completely invisible when it comes time to make the real decisions about a site. Can you answer yes to all of these questions?

• Are the changes my web design team makes actually in response to measured web behavior?
• Has my web measurement ever suggested directions, products, services or systems that significantly changed my business approach?
• Do findings about visitor behavior on the website ever influence other media strategies and messages?
• Does anyone really read or do anything with the web measurement reports they receive?

If not, then join the throngs who aren't getting what should reasonably be expected from analytics.

No excuses
There are many theories in the analytics community about why the answer to these questions is often a big, fat NO,  ranging from poor tools to lack of data to bad marketing organizations. None of these explanations hold up. Though analytics tools have improved dramatically in recent years, there has been no consistent improvement in analysis. More data is always good, but the internet channel is flooded with data and is in no danger of suffering a drought. And while it's convenient to blame the consumer of information, most marketers genuinely thirst for analytic direction.

The deeper problem -- one that's painfully obvious to anyone on the outside who sees the results from most web metrics -- is that most analysis isn't based on any coherent view about how web analytics should be done. Without a consistent, conceptually powerful methodology for doing web measurement, many practitioners flounder.

Functionalism is designed to provide measurement practitioners with a rich, formalized method for doing web analytics. Based on the simple idea that each piece of a website has a particular function and that it's effectiveness in this function can be measured by using statistics that are highly-tailored to its role, functionalism provides a generalized framework for integrating analytics into the business and design process. Here's how-to implement the functionalism methodology:

Classify important website pages
To implement functionalism, you start by classifying the important pages on your site according to their role. This isn't a random process--  the methodology includes a rich set of page types that can be easily identified on most sites: things like router pages (often top-level navigation pages designed to move visitors to appropriate sub-content), convincers (pages getting visitors ready to buy a product/service), closers (designed to get visitors to pull the trigger on a buy), billboards (pages on publishing sites whose primary function is to display advertising)  and re-assurers (designed to provide assurance about some concern during a sales process).

Map key performance indicators
There are many other page types, and the idea is that each page type comes with a measurement specification-- a set of key performance indicators (KPIs) that tell you how well the page is performing its intended function. In most cases, the KPIs are plug-and-play. In other cases, you'll have to figure out how your site maps to the KPIs (this isn't hard-- with router pages, for instance, you have already decided which links on the page are the "intended" routes).

Produce KPIs in the web analytics measurement system
Once you've mapped your KPIs, you'll need to figure out how to produce them with your measurement system. For most of the KPIs this is a snap, but some require a bit of fancy footwork to implement. Naturally, different web analytic tools will provide different levels of support (and paths to implementation). But the KPIs are designed so that most enterprise web analytic systems can handle the vast majority.

Analyze key components
That's about it. You're now in a position to analyze all of the key components on your site in a way that makes sense to every constituency within the web marketing channel. Functionalism provides a framework in which you can agree why a page is being built, what its outcomes should be, how to measure if it's successful and how it compares to other "similar" pages on your site in terms of its effectiveness, so you can decide which ones to try and improve.

After all, isn't this what web analytics is supposed to be for?

Gary Angel is president and chief technology offier at SEMphonic. Read full bio.