Tracking Search Epiphenomena: Why This Behavioral Anomaly is Vital in Web Analytics
For one content area we tested, the average visitor had seven lifetime visits and averaged more than sixteen pages per visit. Both numbers are way above the site average. Is this content area a winner? The question is a trick one — because I haven’t given you any of the essential information to answer it.
Why not? Well, it turns out that in most cases viewing a content area is entirely epiphenomenal to the kind of statistics I’m talking about here. To get to the content area in question, visitors tended to be dramatically more engaged with the site. In other words, the visitors who found the content were the sort that were already quite likely to have many visits and view lots of pages. The content area view didn’t cause the type of visitor — the type of visitor caused the content area view.
In fact, by implying that we found one content area like this I’ve misrepresented the case. Because we found many, many content areas like this – which is part of what helped convince us that the effect was epiphenomenal and not causal. For ad-based sites, smaller, less frequently visited content areas will almost look like big winners when viewed on an average visit/view basis.
So how can ad-based sites understand when an effect is merely epiphenomenal and not central? We’ve found several methods of analysis to be much more fruitful.
First, it’s important to understand that a content area might be successful in more than one way. Some content is inherently single-use – once a visitor has consumed it, there is no reason to expect them to come back. For single-use content, the type of success we’d expect to see is measured most effectively by subsequent visitor returns without viewing the target content.
On the other hand, many kinds of content area are only working well when they are gathering repeat viewers. For this kind of content, the type of success we’d expect to measure is subsequent visitor returns to the original area.
Let’s tackle this second metric first. In a sense, it’s a measure of the mindshare that a particular content area grabs from total site visits. And to measure it, we track how many of a visitor’s visits (and pages) include the target area. By measuring “mindshare” over time, we were able to screen off the vast majority of epiphenomenal effects — and establish a much better measure of how influential a content area actually is on return behavior.
But what about tools (and content) that are inherently single use? For this type of content, we’ve found that the most effective analysis is essentially to construct an artificial A/B test. We construct a population segment of first-time content area users and then create a matching set (across key variables like pages viewed and prior sessions) of visitors who didn’t see the content. By measuring the difference in subsequent over-time return behavior, we have a measure of whether or not the content actually improved (or degraded) visitor loyalty.
It’s unfortunate, but the vast majority of visitor segments and web KPIs are extremely vulnerable to measuring effects that are mere artifacts of the site navigation and the types of visitors that are most likely to stumble on them. This problem is exacerbated by the restrictions that many web analytic tools place on the complexity of the rules that can be used to create a visitor segment and in their limited ability to track the sequence of events over-time.
For ad-based sites, segments that are built off simple rules about viewing content and use common core KPIs like average visits or page views are almost guaranteed to be misleading. Segmentation (like conversion) has often been put forth as the “right” and necessary way to do web analytics. And, of course, it is very important. But just as measuring versus conversion is far more complicated than it might at first seem, so too can segmentation be badly mishandled. Taken simplistically and measured against the wrong type of KPIs, visitor segmentation can steer both analyst and publisher disastrously wrong!
All of which make epiphenomena pretty darn important. So now, I’m going to go check Google and see if I’m ranked ahead of the Wiki…
Reader Comments.
Many electronic publishers (and transactional sites) are struggling to address the issues that you outline, namely:
-Which statistics matter?
-Where are the actual “choices” occuring that can be affected?
-Given competing objectives (page views vs click revenue), how do I optimize?
There is a suprisingly simple answer to these questions. Properly structured tests enable the publisher to;
-Focus attention on the communication attributes that actually impact behaviour.
-Quantify the impact across multiple independent objectives.
-Weight the objectives based on their relative value.
This framework empowers a combined exploration, discovery and optimization cycle that can be used to optimize content across page views and revenue for different customer segments.
We are seeing this used with great effect for leading publishers and e-commerce sites to gain new insights and improve month-month results.
- Pingback from The Way of the Rich Hermit » Blog Archive » Brulant, Inc. Acquires SEO/SEM & Usability Specialists eMergent
Leave a Comment
Pages: « previous page 1 2
No TagsArticle Sponsor
More Features
-
Loading ...
Latest News
- Twitter Knows Where I Am; Gulp March 12th 2010 ADOTAS – I log in to Twitter to share with [...] more »
- WWN Has a Stopwatch on Ad Visibility March 12th 2010 ADOTAS – Considering how much time people spend online, why [...] more »
- Jumptap Offers Self Service Lane March 12th 2010 ADOTAS – Possibly inspired by those neat-o self checkout lanes [...] more »
- AdPix Barges Into the In-Image Market March 12th 2010 ADOTAS – Watch out, Image Space Media — there’s a [...] more »
- Got Funds? Todacell Scores a Cool Million March 11th 2010 ADOTAS – After securing $1 million last summer, Todacell has [...] more »
- A Custom Postcard Thanks to Your Favorite Brand March 11th 2010 ADOTAS – Handwritten letters may have been trounced in the [...] more »
- Display, but Nothing Else, Hops on YouTube Mobile Site March 11th 2010 ADOTAS – Google has officially opened the gate to display [...] more »
Features
- Amazon Leaves Colorado Affiliates Out in the Cold March 12th 2010
- Secret Salsa Recipe: The Best Digital Marketing Mix March 11th 2010
- The Basics of Brand Protection March 10th 2010
- Agencies: Working Hard or Hardly Working? March 9th 2010
- Why Big Brands Will Love the iPad March 8th 2010
Spotlight
AdBidCentral’s CEO, Vivek Veeraraghavan Talks Openly*What was the inspiration to start AdBidCentral? The conditions that inspired AdBidCentral came from a variety of factors in my personal [...] more...
Reader Favorites
Classifieds
- Sr Director, Marketing Services
- Senior Web Analyst
- Account Director
- Director of Analytics
- Manager, Business and Trade Media Relations
Recent Comments
- Bulent: I wonder how many of the clients would accept a media plan, that would -as
- Nickster: How about we just vote out all of these socialist big government politicians and take
- BWB: Why doesn't the legislation target the affiliates directly? Colorado based affiliates are perfectly positioned
- canoe: The law never keeps up with technology but in this case the law already is
