Forum Moderators: phranque
Given NATs, dial up accounts, etc... we usually get less unique IPs than there are actual people, but then you get home machine vs work machine which ends up flipping the ratio to get less people than the unique IPs would suggest.
I know this is a difficult questions to answer and it's not that critical, but the thought crossed my mind and now it's bugging me... I'm hoping someone somewhere has read something that might answer this.
As far as AOL type IPs, if you are viewing a report that breaks down a visitors page visit pattern - then you need visibility of visit time to determine how likely a visit is realy from a single user. The longer a break in time between pages. The more likely it is really a new visitor. And consider how likely you are to get visitors from same ISP in similar site pages in a short period of time.
As for spiders using various IP's. Visibility here depends on assumptions a log analyzer makes in grouping different IPs together. Usually grouping by http referrer will group things together pretty well.
All in all I find the specific IP address information far to difficult to wade through by eyeballing. The better log analyzers understand these issues and present this info as visitor patterns & grouped referrers for spiders/serach engines.
