NONE: re: visits vs. hits
re: visits vs. hits
William Henry (bobhenry_at_cymedia.com)
Mon, 22 Jul 1996 23:58:55 -0400
At 09:27 PM 7/22/96 -0500, you wrote:
>Like everyone else, we're very interested in measuring web traffic and
>have read the discussions about what are relevant numbers and what are
>not. We use a stat program to measures traffic to our site,
>"iisstat.pl". The program generates a report showing total transfers of
>each file, including the number of file sent for each particular html
>page or graphic. My question is, doesn't this measure give you a pretty
>good indication of how many people are viewing your pages? If anyone
>else is using this program and is familiar with the report it generates,
>we would welcome hearing an analysis of what "number of files sent"
>really means.
>
>-----------------------------NEXT
>
>
>Does anyone know of a "quick and dirty" method to calculate a general
>figure for the number of visits to a site based on the number of hits? Is
>there a simple percentage I can use since I have no time for fancy
>algorithms and cannot afford expensive software.
>
>Any advice would be appreciated.
I can attempt to answer both of these questions (since they are basically
the same question)....
I use Accesswatch for daily statistics. What this program does is count
each unique domain name entry in the server log as one visitor. This does
lower your count of unique visitors a little, since most people are on
dynamic ip addresses that more than one person may connect through per day.
there may be 20 accesses by ip25.erols.com, but accesswatch calls that ONE
UNIQUE VISITOR. Maybe 2 people dialed into erols.com and got ip25's modem,
and one read 8 pages, the other 12. It's still ONE unique visitor. I know
this isn't perfect, but it's pretty close. I think the odds of two people
connecting to the same site from the same ip address on the same day are
high. The odds are even higher of 3 or 4 or 5 people doing so. So, the
unique visitor calculation is "good enough" to give you a general idea of
traffic to a site.
This method considers each ip address as a unique visitor for the day, no
matter how many people actually see your site through that ip. Even the
most foolproof method would not know if ten people are looking over the
user's shoulder, so we can't ask for EXACT scientifically proven stats. The
same user becomes another unique visitor again tomorrow, too, if he/she
really likes your site and comes back tomorrow. That presents another
possible "flaw" in the method, but I think that it actually works as an
offset to the first "flaw". The odds are probably inverse to the
multiple-user-on-one-ip flaw, so the stats may end up being accurate in the
end, through one flaw negating the other. (well, it could happen! really!) ;)
Access watch is a nice stat program, if anyone is interested, the url for it is:
http://netpressence.com/accesswatch/
and here is the top section of my results page from accesswatch, from
earlier today...
--
Today, there have been a total of 2083 accesses by 215 unique hosts viewing
an average of 9.3 pages related to Local Music Online. Of these, 0 (0%)
have been from Cybermedia, and 2083 (100%) have been from outside hosts.
There have been a total of 7749 hits and 27 errors related to Local Music
Online, accounting for 100% of total server hits and consisting of 14443
kilobytes of information. There have been 83.2 accesses per hour, and at
this rate, Local Music Online will get 2615 accesses today.
--
(I've also written a "gateway" to accesswatch, so you can run it from a web
page, rather than going into the server's console. Which is nice for when
you are on vacation.) Ask for it nicely, and I'll send you a copy of it.
(it's in perl)
Bob Henry
bobhenry_at_cymedia.com
http://www.cymedia.com/dcmusic/index.htm
Next time you wave, use ALL of your fingers!!
