Forget Votes, How Do You Count Visitors?
from the take-a-wild-guess... dept
In the early days of the web, one of the big issues was how to accurately count visitors (though, idiotically, many people put more emphasis on “hits” over visitors). It appears that issue has never really been solved. At Techdirt, we use two different tools to analyze our visitors — both working off the same log file and they come up with vastly different answers on the “how many unique visitors” question. I haven’t decided whether to just believe one over the other or simply average the two results. In the meantime, I ignore the issue and concentrate, instead, on the trends. It appears that bigger sites with a lot more money than we have are experiencing the same problem, with a slight twist in the form of companies selling reports on how much traffic a website gets. While log files tell one story, reports from Nielsen and Media Metrix tell another, and if one thing’s clear it’s that a site’s log file, Media Metrix report and Nielsen report are unlikely to come anywhere close to agreeing on how many visitors any site has. They don’t agree on methodologies, and the end result is that no one really trusts any of the numbers, but they still buy them. In the article, it’s suggested that the industry boycott both of these providers until they can give more accurate results, but that’s never going to happen. Instead, companies will continue to do what they always do: watch the trends for themselves while cherry picking the most impressive looking numbers for publicity/self-delusional purposes.
Comments on “Forget Votes, How Do You Count Visitors?”
Some visitors are fake
I have a rather popular guestbook on my website (usually two or three entries per day).
The page is loaded over 100,000 times per month though. Most of the traffic is created by spiders searching for email addresses. I can detect about 30 of them, then I just send them to a page full of fake emails.
My point is that you cant even trust unique visitors, because they may not be actual people.
I only worry about bandwidth at this point.
When that weird email scooping virus came out a couple of months ago my bandwidth limit was exceeded (18 Gb over three days). That sucked.
Cache
So long as we have cacheing then you will *never* get an accurate visitor count, especially if that cacheing is working properly. For common sites I bet the requests leave our site about ten times a day at best, but there will have been hundreds of hits.
On the flip side
I’m currently envolved in monitoring visits to a corporate intranet. Every user has to log in and it’s recorded in the web log. I know exactly how many visitors I have 🙂
No Subject Given
My site gets about 250 unique visits a day according to the server side log software my web host provides – and about 450 per day according to the freeware log analyzer I run against the log on my local machine.
I really don’t know which one is closer to correct – although the local software last month reported over 6000 visits due to a trackback to a very popular weblog. That would seen to validate the larger number I think.
Not that it matters – my site is a hobby so there is nothing riding on the traffic – not even my ego 😉
Site Visitors
No mention of visitors via RSS – I’ll visit to comment on a post but I rely on NewGator to keep me up to date about what’s going on where.
Re: Site Visitors
Can’t you track RSS requests from unique IPs, though? Not that it matters for advertising purposes (unless you stuff ads into your RSS feeds!)
I have been using a combination of unique IP address / user agent to determine unique visitors. This compensates for the corporate firewall effect of hundreds of people having the same IP (they rarely all have the same user agent). It also takes spiders out of the equation (unless the spiders are real clever and spoof their user agent).
I’d be interested to hear how you guys determine unique visitors…
The Shared Lie
There’ll NEVER be a boycott. The ad agencies & media buyers are sheep, and will all buy into a shared lie, as long as *everyone* is buying into the same lie – it’s no different than TV Neilsens. Everyone in the Entertainment industry *knows* the methodology is more than suspect and that the number of viewers listed doesn’t reflect reality, but it’s a lie they ALL share; as long as everyone’s using the same measures, the relative success/failure of a given show (website) is accepted.