Functionalism Basics

In this post I'll describe the basic building blocks of functionalism. As you follow along, think about how you can apply these techniques to your own web site measurement and optimization projects.

Functionalism Defined

Functionalism is a methodology for web measurement. It includes:

• A way to understand what each part of a web site is supposed to accomplish.
• A specific set of measurements to determine how well the function is being performed.

Here's the cool part - functionalism is appropriate for any type of site: commerce, lead gen, media, branding, support, etc. 

Back when I was an on-staff web analyst at a large company with a very diverse collection of web sites - none of which had direct and obvious ties to revenue - I used to joke that, "I wish we just sold shoes." No offense to analysts who measure commerce, but you people have it easy. If I had known about functionalism back then, my group's web analytics projects would have been far more structured and consistent than, sadly, they were.  

So much for my flashback. Here in the present, functionalism can give you and your web analytics team a framework for measuring every component of every diverse web site that you own. Not only can it enforce business-wide consistency in basic measurement, it can also put some much-needed structure around the process of doing web analytics. 

The Approach

Functionalism has 2 major components:

• A set of page types that map to content on your site 
• A set of KPIs that map to each of the page types

Once these components are in place you can take any page on your site, give it a page type classification, then measure how well it's performing using the KPIs associated with that page type. In this way, functionalism provides a method for deciding when and how to use any KPI that you define.

Page Type Classification

Page types are designed to capture the natural function of a page. For instance, you'd call a page a "Router" if its function is to move the visitor to specific places on the site, and you'd call a page a "Convincer" if its function is to convince someone to buy your product or service.

As I classify pages I often refer back to a definitive list of the page types in my repertoire. It looks like this:

• Engagers: Pages whose primary purpose is to interest the visitor and get them to do something/anything on the site.
  
• Routers: Pages whose function is to move visitors into specific places on the site. One of the differences between an Engager and a Router page is that the latter is built with the expectation that the visitor has come to the page looking for a particular type of information/service. Search should often be treated as a special class of Router page.
  
• Convincers: Pages whose function is to “sell” the visitor on a product or service.
  
• Explainers: Pages whose job is to help the visitor understand some aspect of a product or service.
  
• Informers: Pages whose primary objective is to provide basic news and information about a product or industry. This differs from Explainers in that content is not specific to your own particular product or service.
  
• Billboards: Pages providing various content to visitors, but whose primary business objective is to display third-party advertisements on a Cost-per-Impression basis.
  
• Sponsors: Pages or series of pages which are themselves revenue-producing, whose content is mostly provided by a third-party who pays for inclusion within your website.
  
• Closers: Pages that are supposed to get visitors to enter a conversion process.
  
• Converters: Pages that are part of whatever is necessary to gather information/agreements and get a finished lead/sale/transaction.
  
• Re-Assurers: Pages built to re-assure the visitor about some potentially problematic issue or concern (privacy policies are a common example).
  
• Tools: Pages that are designed to collect or provide information as part of a nonsales process (like checking an account status or finding a location). There are many different kinds of tools and for a tool-rich site tool pages need to be subdivided into more granular types.
  
• Completers: Thank you pages – designed to signal the completion of a process and – in some cases – drive to additional engagement.

This is the Cliff Notes version of the full page type listing that you can find on page 19 and following in Gary Angel's freely-available Functionalism White Paper.

Make a list like this for yourself. Keep in mind that, as long as you have defined the right page types, every page on your site can be classified. Don't feel bound by the list I've provided here - it's totally fine to create your own page types.

Page Types & Associated KPIs

Within the framework of functionalism, each page type is associated with one or more KPIs. In this way, the measure of a page's effectiveness is tailored to the function it performs.

For instance, a "Router" page will be associated with KPIs that measure how well it moves visitors to specific places on the site. Typical router KPIs will measure the "next page" behavior for that page - the percent of clicks on body content, percent of clicks on navigational links, exit rate, etc.

Similarly, a "Convincer" page will be associated with KPIs that measure how effectively it convinces a visitor to buy. The most common convincer KPI is percent of next page clicks on call-to-action links - you can call this "Conversion Rate" as long as your business audience is clear on the definition.

Here's a sample mapping:

  

As with page types, the KPIs can be customized to suit your needs. Pick KPIs that 1) make sense as a measure of functional performance, 2) are meaningful to your business audience, and 3) are actually obtainable from the web analytics tool that you use.

KPIs are tool-neutral and designed to be implemented in a range of solutions including Omniture, Webtrends, Google Analytics or any other tool. So as long as you've got a basic, valid visitor behavior tracking in place on your web site, you can use functionalism today without having to customize your implementation.

That's the basic setup. Once you've defined how you'll measure each page, you can use it to compare pages of one type to one another, to assess the effectiveness of redesigns, and to provide focus for ongoing optimization. I'll cover examples in future posts.

Functionalism is NOT ...

Now that I've outlined what functionalism is, here's a quick note on what it's not. Functionalism is not designed to provide any subjective analysis of pages: It includes no subjective design rules, no business specific classifications, no way to “score” pages subjectively. 

It simply provides a framework for performance measurement.

[This post is part of my series on Functionalism. In my next installment I will walk you through a how-to.]

Comments

[Les]

Hi June - I had been waiting for this blog post. And then when you finally posted it, I didn't have time to read it - go figure.

Well I made up for my lapse in timing by reading your article twice!

This is an excellent perspective on Analytics. I like the approach - and cannot wait for your next BLOG entry on the topic!

[June Dershewitz]

Hi Les. Thanks for reading my blog! I can tell by checking out *your* blog ( http://www.websitestatistics.ca ) that you're just as passionate about web behavior data analysis as I am. If you wind up using the functional techniques in your own practice, I invite you to post about it and share your perspective. Cheers, June

[Trinadh]

excellent post..!

[Rudi Shumpert]

June,

I enjoyed this post. I missed it when you first posted it. This approach really appeals to the coder in me. It's practical and makes a lot of sense.

Now to go apply this to our site.

-Rudi

[Nancy D]

June,

I really enjoyed your Beyond Web Analytics (@BeyWebAnalytics) interview. Great post and I'll read the white paper too. This will help me focus my analysis work.

-Nancy

[Shilo Jones]

Hey June, since Gary first wrote about Functionalism back in 2006 I guess as of the writing of the White Paper and now fast forward to 2010 and the number of times you've actually used Functionalism in practice, are you finding that the tools are catching up with the concept to make functional analysis easier?

Specifically, we've been starting to make some progress on this type of work doing what we had called Page Type analysis and am now trying to brand as Functionalism on our team. What we've done is to group pages together to aggregate their collective data so we can try to improve the template as a whole for the site and/or have a benchmark to look at high and low performing pages withing a given template i.e. Product Detail Pages or Convincers.

In Google Analytics the only way we could figure out how to do this was to create a separate profile and use regular expressions to classify the thousands of pages we have in to page classifications as you guys called them in the White Paper. It was nice that GA had the feature to allow this to happen, but a bit of a techy way to arrive at something that seems so fundamental to Web Analytics.

Also, some of the calculations on the KPIs you've called out for specific classifications still need to be done outside of the tool itself. Is that the current state in your experience?

Also, I'd be interested to hear maybe in future posts more experiences you've had in the field with Functionalism, the evolving KPI/classification landscape, where it seems to breakdown, etc.

[June Dershewitz]

Shilo, you've asked some *hard* questions! I'm going to answer your techie question here but I am also planning to do a follow-up post or two to address some of your additional requests (hands-on examples, the evolving landscape, where Functionalism breaks down, etc). Thanks for giving me fuel to continue writing on this topic.

Okay, the techie question. If you have not yet read Adam Greco's post on page type pathing, you should definitely check it out: http://www.the-omni-man.com/sitecatalyst/adamgreco/2009/08/02/page-type-pathing/ His example is Omniture-specific, but it addresses some of the concerns that you've mentioned regarding the mass-classification of pages. In Omniture or GA or any other tool you could stuff page type into a custom variable - ideally passed directly from your CMS - and then segment or filter on that variable.

Adam chose not to use the custom variable approach, and instead recommended using either SAINT classification in Discover or DB Vista rules in SiteCatalyst. Unfortunately I don't think GA has an equivalent (yet), so you might experiment with the custom-variable-from-CMS method. Alternatively, you could store the page type lookup in Excel and then use a GA Excel plug-in like Tatvic to generate refreshable reports.

And finally, a word of caution: I've found that Functionalism works best as a focused analysis project rather than an ongoing monitoring system. If you're going to run Functionalism reports continuously on your site, make sure you have an analyst whose attention is devoted to interpreting the results, providing insights, and seeing the recommendations through to completion. Good luck!

[Shilo]

June, thanks for the follow-up post. I'll definitely check out Adam's post on page-type pathing, Tatvic and will certainly look forward to reading your future posts on the topic.

Shilo