NONE: Re: BANNER TRACKING
Re: BANNER TRACKING
eBORcOM (eborcom_at_eborcom.com)
Mon, 22 Jul 1996 11:22:41 +0100
Raymond Lowe wrote:
>In article <199607201351.IAA29731_at_arganet.tenagra.com>, you wrote:
>>>We are currently looking at putting 3 banners on selected sites and whilst I
>>have put the budget together, I am having difficulties in finding how to
>>assess which banner of the 3 is being successful (or not).
>
>Just have the three banners point to different pages on your site! Indeed
>if the banners are placed at different types of sites you have a chance
>to customise the entry for people from those sites appropriately.
This option is by far the easiest way of getting round the problem, but if
you don't use your server's statistics carefully you will not get an
accurate record of how successful your banners and the placement of your
banners has been.
Robots, not only from search engines such as Alta Vista et al, but also the
increasing number of intelligent agents left to surf the web, may decide
that they want to follow the link to your site when they come across your
banner. With intelligent agents this is not so much of a problem, but it
only shows how successful your banner placement has been, it has no bearing
on the quality of your banner. Intelligent agents don't see a banner and
think "that looks interesting, I'll click on that".
Search engine robots are more of a problem. You may find that your new pages
get indexed, and people visit them who have never seen your banner, but who
find the page in a search engine index.
Also, people who click on your banner may bookmark the page that the banner
takes them to. They may, in the future, visit this page from their bookmarks
file. You will have registered 2 responses to your banner when you only got 1.
I suggest that (if possible) you log the following environment variables:
1. HTTP_REFERER - This will tell you which page the user was at when he/she
clicked on the link to your page. If called from a bookmark file or typed in
manually this variable will be empty. Some older browsers do not handle
this, but most do nowadays, and as people move over to newer browsers the
proportion is increasing.
2. HTTP_USER_AGENT - This will tell you what software the user has. If it
says "Mozilla", "Lynx" or whatever you know it's a real person, if it says
"Web Indexer" or something like that then you know it's a robot.
This method still has its flaws, but it should give you a more accurate view
of how your banners have fared than by examining the raw statistics.
Tom Hukins
--
eBORcOM - The home of the Megaphone publicity service, the only way to make
sure the world knows about your web pages.
WWW <URL:http://www.eborcom.com/>
E-Mail eborcom_at_eborcom.com