__/ [Dustin] on Friday 02 September 2005 12:51 \__
> Roy Schestowitz wrote:
>> Feeds are an excellent place to start based on my own experience. When I
>> decided to set up a news section for computer vision back in May (one of
>> these ideas that pops up in your head while in the bathroom), I knew
>> there was no similar site that provided an RSS feed, not even the
>> long-established Computer Vision Homepage, which was static. So, I set up
>> PHP-Nuke and some months later I have 14 fully-subscribed members and
>> over 100 RSS feed subscribers. I also have some good corresponding SERP's
>> that are sure to have the figures improved.
>> I set up another such site which covers one of my research project. I
>> thought there was absolutely no way that anyone would show an interest in
>> it. Nonetheless, it helped me keep track of my own progress. As it's the
>> beginning of the month, the visual logs are rather clean. This morning I
>> could see that the feed of that site was requested 34 times yesterday,
>> only a couple of months after it was set up.
>> In summary, offer something that others do not have. People are going Ga
>> Ga after feeds at the moment. Not all people have reached the point of
>> information overload, yet. Take advantage of that.
>> BOKMA: Your advice on RSS feeds was excellent. The requests appeared to
>> have nearly doubled in just 2 months.
> I set up feeds for Nerdlance and I notice they get pretty good hits. How
> can you tell though who is hitting them. Is it just search engines, is
> it all the times one person's feed reader refreshes, all of the above...
Putting aside the ability to track unique IP addresses, here is what you can
do to get a rough guess:
o Wait until enough hits accumulate, e.g. one month
o Take the number of services you ping into consideration
o Also consider how often you ping the service
o Assume that most people read (refresh) their feeds once or twice a day
(I would love to find some studies about the average number though)
o Do some rough math in your head
o Since pinging services are less predictable, you can get good estimates
when your statistical sample (i.e. human subscriber) is large enough. Once
humans dominate, you can just divide the total number of hits by the number
of requests per day, divided by the total number of days.
This still gives you a very _rough_ figure, but it is trends you should care
about the most. For example, see if the total number of hits increases
every month and if so, by how much? Also keep an eye on UIP figures.
Numbers that rely on feeds have a low UIP/visits ratio.
Roy S. Schestowitz | "Black holes are where God is divided by zero"
http://Schestowitz.com | SuSE Linux | PGP-Key: 74572E8E
1:45pm up 9 days 2:53, 3 users, load average: 1.54, 1.03, 0.85