Analyzing
Web Traffic
Gathering information on web site traffic is a crucial element
in the planning of marketing and further development. When analyzing
web site traffic, it is also important to be able to understand
the associated statistics terminology.
A
question that webmasters are often asked is "how many hits
does your site get?". You then need to ask "do you
mean hits, requests, unique visitors or page views?".
Many
hit counters, such as the FrontPage 2000 component can be a
little misleading and display grossly inflated statistics. This
particular hit counter only measures page views and if you sit
on a page clicking the "refresh" button, it will increment.
As
FrontPage and similar counters only reports numbers, it is a
very poor method of gaining an accurate view of traffic and
the way visitors interact with your site - you'll need a more
comprehensive tool.
The
following is an outline of different methods of statistics collection,
plus definitions of associated statistics terminology to help
in analyzing web traffic:
Raw server logs - manual statistics gathering
One
of the best methods of studying traffic, particularly if you
wish to really narrow down on a certain page, browser, IP address
or search engine referrals is to download your server logs.
A server log is basically just a text file that contains every
request that has been made to your server. A request may be
for a page, an image or any other element contained on your
web site.
A
server log entry looks something like this:
216.154.251.84
- - [14/Aug/2005:08:04:14 -0800] "GET /animate.js HTTP/1.1"
"http://www.sfldfkfdk.com/" "Mozilla/4.0 (compatible;
MSIE 6.0; Windows NT 5.1)"
It
all looks a little confusing, but it's really pretty simple
when you break it down:
216.154.XXX.XX
- This is the IP address of the person requesting an element
from your site (x's will display as numbers). You can free tools
from sites such as DNS Stuff to then translate those numbers
to a geographical point.
[14/Aug/2005:08:04:14
-0800] - The date and time of the request
"GET
/index.htm HTTP/1.1" - The file being requested and the
method of transfer.
"http://www.sfldfkfdk.com/"
- Where the request came from. This may be another web site
or a search engine. If it was a search engine referral, it's
most likely that the search string will also be included.
"Mozilla/4.0
(compatible; MSIE 6.0; Windows NT 5.1)" - The user agent
(browser or search engine robot) and operating system of the
person/computer requesting the file.
The
manual study of server logs can reveal many interesting trends
and habits, but it's also very time consuming. Each visitor
to your site may generate hundreds, perhaps thousands of requests
on each visit depending on the complexity of your site.
Manually
dissecting your server logs is great for in-depth studies, but
you'll also want a tool for "at a glance" statistics.
There are many server log interpretation tools on the market,
but the good ones are generally very expensive. Many hosts include
statistics tracking with their hosting packages, but I have
found many to be inaccurate or don't update often enough. A
third party service is also useful as a comparison tool.
Also,
while you'll have access to logs if you are commercially hosted
(ask your hosting service about their location), webmasters
hosted on free servers usually won't have this privilege.
If
you don't have access to your server logs, or are looking for
to complement your own server log analysis, there are many remotely
hosted services available. Remotely hosted services are usually
very simple to set up - simply copy and paste a few lines of
code into your pages. Learn more about these services.
Web
site statistics terminology
Webmasters
and site owners tend to use different terminology in relation
to statistics and it's important to be able to differentiate
between the terms - especially if you are wanting to attract
advertisers to your site. Savvy advertisers will want to know
exactly what your traffic rates are, where it comes from, what
your visitors come to your site for and how long they stay.
What
is a hit or request?
A
hit is the result of a file being requested and served from
your web site. This can be a html document, an image file, an
audio track etc. Web pages that contain a large number of elements
will return high hit scores. Hits are of very little consequence
when analyzing your visitor demographics.
What
is a page view?
A
page view means just that. Once again, it is not a true indication
of how many different people are visiting your web site, but
it is a good way to judge how "sticky" (the ability
to retain the interest of visitors) your web site is and is
an important consideration regarding the possibility of attracting
high paying advertising.
What
is a unique visitor?
A
unique visitor (combined with page view data) is where stats
really count. It is someone with a unique IP address who is
entering a Web site for the first time that day (or some other
specified period). When you log onto the Internet, you are assigned
a unique IP address, or if you are a cable modem user your IP
address is usually "static", meaning that it never
changes.
Your
IP address is an identifier and while you are using it, no one
else on the Internet can utilize that particular set of numbers.
Your number is counted once, usually for a 2-24 hour period,
dependent upon the tracking software.
So
no matter how many times a visitor refreshes or navigates through
your web site, they will only be counted once for the specified
time period. This is by far the more accurate way of analyzing
web site performance.
Note:
the notion of a "unique" visitor as described above
can be somewhat misleading in the instances of multiple users
sharing the same IP. For example, the proliferation of home
networks in recent years may see you having 2 or more people
from the same network accessing your site via the same originating
IP. This can also occur in the cases of universities accessing
your site. In these circumstances, it is more difficult to tell
if 1 person from the same IP is viewing 50 pages, or if it's
10 people.
If
you really need to break down figures from these sources further,
you'll need to do some more detailed log analysis; comparing
items such as access times, browser/OS configuration etc.
The
other way to get a *reasonably* accurate visitor count is to
look at the overall page view average trend for your site. For
example, if the page view average for your site is 3 pages per
visit, and you have one IP showing a page view count of 200
for a period, then you could divide it by 3 to give you a "guesstimate"
of how many visitors originated from that IP.
What
is a referrer?
A
referrer is simply the origins of the visitor to your site,
i.e. the last site visited and the page on that web site.
Gauging
web site traffic
When you divide the number of page views by the number of unique
visitors, this can give an excellent indication of whether traffic
is transient or is staying on your site. If the average is one
page or under, you can be pretty sure that there is something
on your pages that is scaring people off.
Perhaps
the load time is too slow or your landing page isn't well structured.
Remember that due to bandwidth considerations, those first few
elements that display as your page is loading may be the deciding
factor as to whether a visitor waits around for the entire page
to load.
Analyzing
web traffic can take up a fair amount of time, but it is definitely
worth it.
Further
learning resources:
Free
web site traffic reporting services
Identifying
search engine spiders
Creating
effective landing pages
Article
by Michael Bloch
Taming
the Beast
Tutorials, web content, tools and software.
Web Marketing, Internet Development & Ecommerce Resources