 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Site Log Analysis Case
Study
Lessons Learned
Published Spring 2001
An
Interesting Finding
One of the more interesting findings related to the intersection between navigation, user interface design, and content organization. One of the main content areas featured interesting health information concerning the product, natural foods, and science. Unfortunately, the only way to get off the main page of this area to read one of the handful of articles was by selecting a title from a drop-down menu obscurely located at the bottom of the page. Is it any wonder the page views for the second most popular page in the area fell through the floor by 87%?
Lessons
Learned
1.
User Interface Choices Affect Navigation Success
As the above example illustrates, no matter how motivated users are to
learn more about your products, they can't click through if they can't
find the links to get there or if they're unfamiliar with the controls.
2.
Build On Successes
This site featured a popular quiz about the product and related lifestyle
issues. A printable coupon also proved popular. Yet the client was considering
killing these traffic-driving, customer-satisfying features. Instead,
I recommended that the client transform the high traffic quiz results
page and main coupon info pages into "mini-portal" pages. These
pages could earn their keep by driving traffic into highlighted parts
of the site, thus increasing both clickthrough rates and product awareness.
3.
Know Your Referrers, Part 1
Speaking of that coupon, I discovered that numerous coupon aggregation
sites linked to our client. While some companies discourage deep linking
(bypassing the home page and linking directly to a page deep in the site),
I recommended that the client work with the referring sites. Instead of
deep linking to the coupon print out page, the client should request the
referring sites link to the main coupon entry page, or "mini-portal"
page described above.
4.
Know Your Referrers, Part 2
Looking closely at the log reports, I could see that some of the most
popular referrers were certain women-oriented, healthy lifestyle sites.
Come to find out, the client's parent company had advertising agreements
with the lifestyle sites' own parent companies. So I recommended that
the client increase its banner advertising exposure on those sites. Unfortunately,
the nature of the advertising agreements didn't allow such fine tuning
between the client's parent and its partners. Our client was unable to
increase spending on advertising that we knew actually paid off.
5.
Demand Decipherable, Meaningful Reports from Partners
To make matters worse, the periodic banner presentation and clickthrough
reports that the large, monolithic portals and other advertising partners
provided our client were comletely indecipherable. Tried as I might, I
couldn't make heads or tales of them, so I knew they were meaningless
to our client team colleagues. In today's web advertising environment,
spoils should go to the advertising partners who provide the most value
added reports which help the client understand the business return
on the relationship.
6.
Avoid Frames Like the Plague
Frame-based sites suffer from well known usability issues. To make matters
worse, companies which insist on building frame-based sites almost universally
implement the concept in the shoddiest manner. Now there is another reason
to stay away from frames: They complicate server analysis to such a degree
that it's nearly impossible to filter out the "noise" they generate
in log reports.
7.
Plan for Server Log Reporting From the Outset
Make backend technology decisions which support your metrics. For example,
leverage cookies or user authentications to track pageviews. If you build
a dynamic site, make sure the server can still track which content was
served to users. The complicated CGI content serving system we inherited
from the client's former vendor, unfortunately, prevented the server from
tracking many of the pages that were served.
8.
Know the Limitations of Server Log Reporting Tools
Tools like WebTrends provide a lot of fascinating data which can be very
useful for improving the user experience of a site. However because of
some basic design limitations built into the Internet protocols themselves,
site usage reports can be very misleading. Become familiar with exactly
what constitutes a "unique visitor" and a "pageview,"
for example. Also, be aware how caching and proxy servers (like those
serving AOL's 22 million users) can skew your reports.
Other Links in this Case Study
- Case Study Introduction
- Lessons Learned (this page)
- Tips for Customizing Server Log Report Tools