Is Your Funnel Clogged with Useless Data?

By John Marshall

If you're using any sophisticated web analytics program, you're probably familiar with a funnel report. The funnel, simply put, is a stage-by-stage view of visitors traversing your site, typically organized from top to bottom. Visitors from campaigns and other sources arrive at the top, pass through a 'qualification' stage where they're educated on your products and services, and then eventually continue on to your goal page.

Linear Thinking is a Direct Route to Nowhere
At first glance, the funnel shape appears to represent real visitor activity on the site. For example, the funnel indicates abandonment from each stage in the customers' 'path' through the site. And the idea, of course, in funnel analysis is to try to cut down on the number of exits from each stage, until visitors reach the final goal stage.

But this can be an overly simplistic approach. A funnel implies linearity—groups of visitors moving directly from point A to point B within your site. In reality, your web site is much more complex. People are free to click around, revisiting pages they've already seen or even clicking around multiple pages within an individual stage. Somehow—while clicking around all of those linked pages and traversing the site in a very human, nonlinear fashion—the customer becomes persuaded to want to do business with your company.

Typically, when marketers first delve into using a funnel as part of their web analytics, they unknowingly do so very superficially—they view the challenge and the purpose of analyzing their site as being to either 1. Expand the number of people coming into the funnel, or 2. Reduce the number of people leaving the funnel before they make a transaction. Based on that logic, very wide funnels that end with a very narrow conclusion point are viewed as being bad, and cylindrical or more uniformly shaped funnels are seen as good.

Traditional Funnel Conundrums
The first problem with the traditional view of funnel analysis is that each stage of the funnel is typically seen as a single page. This really doesn't reflect—nor can it reflect—the complexities of any moderately successful web site. If you attempt to view your funnel as a single page for each distinct element, the data become too large to interpret.

It's not a computation problem; it's a problem of how much data analysts can fit into their heads. Large numbers of pages need to be reduced down to some kind of logical entity. For example, you don't want to see your entire catalog of products as 10,000 distinct entities inside your funnel. Ideally, you want to see those as a group of pages because they all serve the same purpose in the same way.

The second problem, which is related to the first, is the fallacy of people moving through the site linearly. Traditional funnel analysis is very prone to drawing the analyst into this trap—and it really is a trap. Your site is not navigated linearly—people don't file into the home page like an army of ants, marching from start to finish, with a certain number defecting along the way. They explore. They hit the 'back' button. They follow links. They look around it in a human way.

This Mouse Was Made for Clicking
Your visitors will click around on different content and view pages in seemingly random order. They may put something into the cart and then decide, "No, I don't want to do that. I want to go back to the catalog; I forgot something." The whole concept that your funnel is operating like a machine—with people coming in at the top and either forcefully or willingly being ejected from the site on their way to the next stage—is simply wrong.

Related to that is the idea that there is a path through the site—a path through the funnel. This idea too has pulled the wool over the eyes of more than one marketer. There is no distinct way that people go through this process. If you think there is, do a quick reality check with some usability analysis—we think you'll find we're right on target.

The most significant challenge in traditional funnel analysis is the lack of filtering or segmentation. In all forms of web analytics, segmentation is necessary essential. The whole gamut of human behavior is very different depending on the segment users are in. Are they a user from an e-mail campaign, or did they come to the site from a paid search listing? These types of segmentations produce substantially different behaviors.

It's pretty obvious that someone arriving from an e-mail campaign will behave quite differently from somebody who was referred by a paid search listing. Therefore, analyzing how people move through the funnel must also be segmented. If you don't segment that, looking instead at the 'average visitor' then your data isn't as useful as it could be.

Page Persuasion is the Answer
By now, you probably know that ClickTracks is founded on making data useful—and that's why we've created a funnel report that can be segmented, and focuses on each page group's effectiveness in persuading a visitor to move closer to your goal. Anyone familiar with the work of Future Now, Inc and the Eisenberg brothers will know what I'm talking about (if you don't, check out the links below). Our funnel report is available in ClickTracks Professional, ClickTracks JDC and the ClickTracks Agency Edition and I invite you to give it a try. The results may surprise you.

For those of you new to this idea, here are some resources for more information on our funnel analysis approach and the concept of visitor persuasion:

• ClickTracks Funnel Report Overview
• My Virtual Class on Funnel Analysis
• Future Now, Inc.
• Bryan Eisenberg's Virtual Class on Persuasion
• Use our Funnel Report Free Once a Month