Welcome to WebmasterWorld Guest from 188.8.131.52
Forum Moderators: phranque
"Hits" is a useless statistic though it is used by many to promote sites. A hit is counted every time one file is accessed on your server, whether it be a page, an image, a program, js file, css file etc. Complex or highly visual pages with lots of js or small images therefore will get 40 "hits" for a page with say a css, 2 js's and 36 images. A page with no images or scripts, gets "1" hit.
The "log lingo" really depends on the logging package you are using. Most give good help files. General definitions are poor because there is limited agreement over how to define things like unique visitors, visits, etc.
A "visit" depends on how your web log program defines it, but I'm fairly sure your program is referring to one IP that visits you site. Whether they visit 1 page or many, they are just counted as 1 visit.
A "page view" is a useful index. Just looking at how many times people visited an actual page.
"Unique visitors" is another way to measure. That is also defined by your sofwtare but is meant to count a visitor only once if they visit multiple times. Some popular software defines a unique visitor as an IP that visits more than say 30 minutes apart. They are therefore assumed to be different as an IP visit may in fact be a different person accessing from the same IP.
Don't overinterpret your logs. It is really hard to know what a visitor is due to the fact that logs only log computers that visit your site - not human beings! Some ISP's alocated different IP's every time a dial up user connects etc...
We use logs a lot, but not in an absolute sense. We use them mainly to find referring pages/sites, SE hits, in a relative sense to see how each page is performing over time, and what pages are getting more hits than others.
There are better ways to measure web site performance such as number of subscriptions to your newsletter, number of signs up, number of sales, number of bookmarks, number of emails you get from the site, number of guest book sign ups etc. etc. etc. These are what we call "outcome variables" rather than "process variables" which your log represents.
Also don't forget that not every page view is an actual read by a human being. Most of our page hits are from search engine spiders, email harvesters, page change checkers, and heaps of automated services where the page is not actually "read" by a human. Yet another reason why you should ignore boasts of "hits" that websites claim in their advertising spiels.
Of course using cookies and other technology can give you a better idea of visits.
(Edited for grammar and spelling since Brett made this a hot topic :))
If you click/visit/download/etc. 10 pages in 5 minutes, that is considered one visit. On the other hand if you delay between the transfer of pages 10 to 15 minutes (and this time delay is dependent of your log analyzing software), it is considered separate visits.
I use hits very extensively with the file that was transfered to track and estimate my bandwidth needs.
Hits also tell me if someone is trying to mess with my site. Repeated hits in fast successions, to the same login page would be a clear indication that someone is trying to break in. Most humans give up after three or four tries of failed login.
It gets more complicated when some ISPs use load-balancing transparent web proxy that might appear as different IP addresses on your logs, even all of them might be generated from the same person.
It's an interesting exercise. Uniques, page views, visits, visitors -- even if you could custom define exactly what your stats measure, what exactly do you want to measure? It seems like a clear-cut question: how many people came to my site last week. But answering it is like trying to pick up a drop of mercury.
Is 30 minutes a reasonable cut-off for declaring that the same IP address is actually a "new" visitor?
If you use code that automatically refreshes the page on a window resize, should that count as 2 page views? What about someone whose download stalls so they reload?
Is there a reasonable guess to use for AOL, Deutsche Telekom, etc when they employ up to 7 IP addresses to give a visitor one page? Is it worth the trouble to track those multi-IP combinations throughout one "visit"?
What happens if traffic is really high and more than one visitor is using the same IP?
The closer I look at these numbers, the more all meaning seems to vanish. Ever seen the movie Blow-Up by Anotonioni?