Forum Moderators: DixonJones
Last night I was experimenting with a new tracking script that I installed on my website. I was using a dial-up internet account to test my script, and I was dividing the connections between two local access numbers for the ISP, hoping to create a new "unique" visitor to my site with each connection since the ISP assigns ip addresses dynamically.
For example, I would call ISP phone #1, run my script, then disconnect. Then call ISP phone #2, run my script, then disconnect. Then repeat process again. I completed this process for a total of 10 connections.
Each time I connected to my ISP, regardless of the phone number, I would receive a new (dynamic) IP address. Some of the addresses I received when I would connect were:
67.31.144.158
67.31.176.135
63.183.209.55
63.191.17.183
However, when evaluating my log files, I did not have 10 unique visitors. Instead, my unique visitor count only increased by one or two, although my "total visitor" count went up my 10, as expected. I was using awstats to analyze my logs. I was shocked. I thought that each of the ip addresses (like the 4 above), were considered unique, and would raise my unique visitor count so I could evaluate my new script.
What's going on here? What is a unique visitor? Why doesn't AwStats tell me each connection from each unique ip address is a unique visitor? Are these dynamic IP addresses from my ISP really *NOT* unique?
What I would imagine happened was that you got some pages from a proxy-cache, and so your logs (and server) never even saw the visits.
I read somewhere that there is a way around the proxy caching issue, but for the life of me I can't remember what it was. Hopefully someone else will chime in because I need to remember!
although my "total visitor" count went up my 10
Whoops, missed that. So, scratch my theory!
It excludes certain robots and probably many scripts as well. No stats program is foolproof. You need to read the documentation to figue out exactly what you are looking for.
but awstats is designed to detect unique VISITORS versus unique IPs.
Well, this is exactly where I am confused. I was hoping to create the impression in my log files that I was a unique VISITOR by using a uniquie IP each time I connected to my website. If Awstats doesn't determine unique visitors by IP, then how does it? I'm almost positive that it doesn't plant a cookie, and log files and cookies are the only way I know of to determine visitors (unique or otherwise).
What I would imagine happened was that you got some pages from a proxy-cache
Yes. I have been reading about this. As I understand it, some ISPs (such as the national ISP I was experimenting with), use server side caches, so when a visitor loads a particular webpage, and then shortly thereafter another visitor loads the same webpage, rather than pull the page from the particular website, the second visitor actually pulls the page from the ISPs server cache.
I would like to find out more about this, to determine if somehow this could have an effect on distinguishing visitors vs. unique visitors.
It certainly does. Accurately sessionizing visitors without using cookies is very difficult. If you were conducting your test from AOL (you were rather vague about that) you are demonstrating to yourself the impossibilty of it.
The problem with proxy servers is that a single AOL visitor in your logs may display a different IP address for every page view. So what looks to be 10 visitors who each looked at one page each, may actually be one AOL visitor who viewed 10 pages. Quite a discrepancy.
Log files are good for determining what your server is doing and what requests are being made of it. They are not particularly acccurate for visitor tracking.
The problem with proxy servers is that a single AOL visitor in your logs may display a different IP address for every page view. So what looks to be 10 visitors who each looked at one page each, may actually be one AOL visitor who viewed 10 pages.
Right. I understand that. But, that's not what the results showed. My logs showed the opposite. One unique visitor, and 10 visits. If my ISP was doing this server cache thing, wouldn't it have showed more like 10 unique visitors multipled by the number of pages they viewed (in theory). OR, one unique visitor and one visitor (ie., if all subsequent pages were pulled from the ISP after my initial visit)?
BTW, the "national ISP" used was Earthlink. :)
Has anyone else looked at referal strings to identify unique visitors, or does any software (combined with other factors of course)? This is something I plan on experimenting with for my own custom stats. Right now I'm using IP+agent, but I keep looking at that referal data and "seeing" the unique visitors in that.