Technology Corner: Web Site Stats - A Look Behind The Numbers
By Joe Gannon
In the dot.com boom of the 1990s, an electronic gold rush began as companies
flocked like new-age prospectors seeking to plant their stake in this digital
revolution that has today transformed the ways companies communicate and
do business around the globe.
But building a web presence isn't easy - it requires a sizable investment
in technology and resources. Forrester Research estimates that building
a typical content web site costs US$420,000, with first year maintenance
costs of US$892,000. That's why, having a web presence today isn't
good enough - there needs to be a way to analyze the return on investment
from using this technology.
This article will focus on how web site content can be measured
to determine that critical question. By better understanding what
types of measurement methods are available, communications professionals
can make informed decisions about how well their content is really
doing.
Server Log Analysis
Typical measurement of web traffic involves analyzing log files
produced by the web server. These logs can contain a huge volume
of data, and may appear cryptic to the untrained eye. While numerous
web analysis software packages can make the numbers easy to digest,
it's still a good idea to first understand how they're produced.
Here's an sample of a fictitious request produced by a user during
a visit to a site's home page:
Reading from the left, this shows:
1. The request date and time in Greenwich Mean Time. The (-0700) means the
server is 7 hours behind GMT.
2. GET -- tells the web server the browser is requesting a document, or program.
3. The name of the page requested.
4. The request is being sent using the Hypertext Transport Protocol. (HTTP).
5. Status Field Code - Describes whether the request was successful. A 200
indicates OK.
6. File size being transferred from the server to the user's web browser, in
number of bytes.
In the above example, the page request generated 4 requests, or "hits." That's
why hits have no direct relationship with the number of pages downloaded.
Instead, page requests (also known as a page impression or view.
) is more commonly used.
But measuring page views be can be tricky. Do you count it when
the page is requested, or fully loaded? Also, you have to be careful
about external factors:
- Search engine robots--which index web sites--can increase your
numbers.
- Caching, which stores files on the user's computer after an initial
request, undercounts your numbers.
- It's difficult to determine whether a user really "viewed" a
web page.
That's why some organizations such as the Interactive
Advertising Bureau are trying to come up with more standard
measures. For example, a page impression is defined as "responses
from a Web server to a page request from the user's browser, which
is filtered from robotic activity and error codes, and is recorded
at a point as close as possible to the opportunity to see the page
by the user."
Performing more complex analysis regarding specific user behavior
may require registration or cookies.
Measuring Value
Putting a dollar sign behind your numbers can give you a better idea of the
content's bottom-line impact.
There are several ways to measure it:
- How much money did you save from putting the content online?
Compare the printing costs with the cost of producing it online.
- How much time did the user save from viewing the page, compared
to the alternative action?
For example, an online meeting can reduce travel costs. The user
also benefits since they save time by not having to travel.
Additional Considerations
Before looking at the numbers, communications professionals consider
the following when putting content online:
1. Findable. How easily can a user find the document? Users who can't
easily find it may give up, thus reducing usage.
2. Purposeful. What's the content's overall purpose? For example increase
brand awareness, or announcing a new product. There may be multiple purposes.
One way to measure whether the content meets users needs is to add a short
survey at the bottom of the article.
3. Actionable. When a user clicks a piece of content, they should be
able to perform a specific actionable task, such as order a product, request
product literature, sign up for a webcast, or subscribe to product newsletter.
4. Usable. A document with poor information architecture or structure,
labeling, or organization, will negatively impact usage.
5. Linkable. One of the web's most powerful features is linking to other
documents. This enhances the user's opportunity to get more information than
would be possible by reading a typical paper document.
Finally, remember that web logs are only one measurement tool. For example,
user feedback can provide valuable information to help you better understand
how well your site is doing.
Here are some related resources for more information:
Server Log Analysis Tools
http://dir.yahoo.com/Computers_and_Internet/Software/Internet/World_Wide_Web
http://www.webtrends.com/
http://www-3.ibm.com/software/tivoli/products/web-site-analyzer/
http://www.sas.com/solutions/intellivisor/
http://www.spss.com/pwa/
Search Engine Opitimization
http://www.web-ignite.com/
http://www.positionagent.com
Web Usability
http://www.useit.com/
http://www.uie.com/
Cookies
http://www.cookiecentral.com
Books
Web Metrics -- Proven Methods For Measuring Web Site Success,
Jim Sterne
(Wiley Publishing)
Web Site Stats -- Rick Stout (Osborne/McGraw Hill)
Joseph Gannon has been involved in electronic and Internet publishing
for 7 years with Crain Communications, Online Focus and Gas Research
Institute. He currently is a web production manager with the American
Veterinary Medical Association, and publishes a bi-monthly web
design magazine http://www.ganemanrussell.com.
Discuss this topic with other IABC members at: www.iabc.com/memberspeak.
|
