Reasons for Discrepancies in Click-Through Rates

From: Bryce Isham <bisham_at_claritas.com>
Date: Wed 17 Nov 1999 10:21:54 -0500

A while ago, I posed the question of why click-through
rates recorded by web sites that hosted our banner ads
differed from click-through rates (measured by user
sessions) in WebTrends reports that we ran. The
banner-hosting web sites always reported about twice as
many click-throughs as our WebTrends reports showed.

A number of people provided me with possible reasons,
plus I checked out the WebTrends portal site, checked
out some issues with WebTrends technical support, and
thought through possible reasons. Below are the
reasons why we could have been experiencing the
discrepancy in click-through numbers. All comments and
additions are welcome. And thank you to everyone who
helped with this list of reasons!

Reason #1
=========
The page(s) to which our banner ads have been directed
loads too slowly.

It is likely that a number of people who clicked on our
banner ads canceled the page load before reaching our
web server.

The results:
-- one click-through on the report from the
   banner-hosting web site
-- zero click-throughs on the WebTrends report

Solution

This problem can be eliminated by creating a
quick-loading page -- 10 seconds maximum for loading --
for site visitors that come via a banner ad.

Reason #2
=========
The banner ad click-throughs could be URL cache-able.
Proxy servers of national ISPs and large corporations
may be causing click-throughs to be directed to their
previously cached documents.

Let's take the case of a national ISP. Within a certain
time period determined by the ISP, one user could click
on the banner ad and get to our web site. Within the
same time period, another user visits the same
banner-hosting site and clicks on the banner ad -- but
this time our web page is pulled from the ISP cache,
not our web server.

Pulling the document from the ISP cache generally will
not result in an entry in our server log files.
Sometimes, however, it will result in an entry in the
server log files, but the entry will contain no
referring URL information.

The results:
-- two click-throughs on the banner-hosting web
   site report
-- one click-through on the WebTrends report

Solution

Bust the cache! Cache-busting techniques are available.

Reason #3
=========
Two visitors with the same IP address count as one
click-through on the WebTrends reports.

If two visitors with the same IP address click on our
banner ad within the same time period, the WebTrends
weekly report will show these two visitors as one
click-through.

The results:
-- two click-throughs on the banner-hosting web
   site report
-- one click-through on the WebTrends report

Solution

A cookie plug-in should solve this issue.

Reason #4
=========
Some visitors to the banner-hosting web site will
double-click a banner ad.

The results:
-- two click-throughs on the banner-hosting web site
   report
-- one click-through on the WebTrends report

Solution

Contact the banner-hosting web site or advertising
service that manages the banner ads) and see if they
can filter out the double click-throughs.

Reason #5
=========
Our company employees visit the banner-hosting web site
and and generate click-throughs.

Click-throughs from our employees are included in the
click-through reports from the banner-hosting web
sites, but are filtered out of the WebTrends reports.

The results:
-- lots of click-throughs on the banner-hosting web
   site report
-- zero click-throughs on the WebTrends report

Solution

You can only determine the extent of this problem. Run
the WebTrends reports with and without filters to
determine the impact of these employee click-throughs.

Reason #6
=========
If not properly set up, WebTrends reports do not cover
the full week of activity.

The results:
-- more click-throughs on the banner-hosting web site
   report
-- less click-throughs on the Webtrends report

Solution

Realize that the raw log files are in Greenwich Mean
Time, and pull appropriate log files to take this time
factor into account.

Reason #7
=========
And more!

One example is internal redirects.

A particular banner-hosting web site could be using
internal redirects. This means that when a visitor
clicks on our banner ad, she is instantaneously and
invisibly redirected, via a script, to another page
which she never sees.

In this situation, it is possible that each
click-through is processed and logged twice on the
banner-hosting web site -- once when the banner ad is
clicked on, and once when the visitor is sent from the
internal redirect page to our web site.

The results:
-- two click-throughs on the banner-hosting web site
   report
-- one click-through on the WebTrends report

Solution

In the WebTrends reports, look only at the
click-throughs from the redirect page.

Thanks.

Sincerely,
Bryce Isham
Product Manager, E-Commerce Group
Claritas Inc.





Received on Wed Nov 17 1999 - 09:21:54 CST