Server based statistics should be regarded as inaccurate or at best a rough guide.
When an individual accesses a web site the server logs:
IP (provided by their IAP)
Internet Access Provider 
Length of Time on Site
Where the visitor came from to get to site
Where the visitor went when they left site
Reasons for inaccuracy:
Picture a large corporate office with 100's of computers, all hooked up to a router which connects all the systems to the Internet. This router has been assigned a few IP addresses used to represent all those systems. When people from that network access a site, the server logs record the IP of that computer which was assigned an IP from the shared IP Pool, which the Office Router is in charge of. Meaning if 10 individual computers within that network access the web site, the server would likely record the same IP, causing the statistics program to display 1 unique web visitor and 10 impressions. The IP address belongs to the IAP , therefore any server can resolve what IAP the site visitor uses. Live in AZ but surf the net from VA? Geographic statistics are derived from your IP. In Alberta one of the primary IAP's is TELUS.net. When a TELUS Internet User accesses a web site, the geographic report will state the user came from Alberta. When a WHOIS is performed on the domain TELUS.net, Alberta is resolved as the location of the user accessing the web site. This is because TELUS' domain, TELUS.net is registered to their Alberta offices. The issue is TELUS also provides Internet access in Ontario. When an Ontario TELUS Internet User accesses a web site, the geographic report will still state the visitor came from Alberta because according to the WHOIS , TELUS.net is located in Alberta. This really becomes a catastrophe when Internet Users whom use AOL visit your web site. Your geographic statistics will claim that all your visitors are from Alexandria, VA when in truth they could be from Pheonix, AZ and any other city in North America were AOL provides service. Does it filter? Code Red visits more sites than all Internet Users combined. Code Red is a virus that infects a web server and then gets the infected web server to indefinitely search for other vulnerable servers by visiting them through web sites the server hosts. If a server isn't properly secured, Code Red will happily infect it. How this affects statistics is the server's logging system records the visit just as any other visit since the virus uses someone's legitimate computer, which has an IP and an IAP , all that is needed. The problem this creates is illegitimate traffic reports. Fortunately, there are statistic programs available which can filter the illegitimate traffic based on a command the virus attempts to execute when visiting. Nonetheless there are other programs & robots available which attempt to exploit servers, whose commands are more sporadic, making filtration difficult. Caught With Your Hand In the Cookie Jar? Some suggest cookies are the solution. Getting the web site visitor to enter specific information about them self, recorded to a file and stored on the users' personal computer is the "cookie". When the visitor returns to the web site their information is logged by the web server. While repeat visits may be tracked with increased accuracy, some issues are raised. First being privacy, many schools have strict policies against recording of such information. Second, accuracy of the information being filled in. It is fully up to the user to supply valid information.
|
