I received an email the other day from a colleague asking the following:
“I was surfing for info on Unique Users and came across this – I’d really appreciate your views on it if you’ve got time to have a look. Do you know anything about this company and would it be worth getting involved with them?”
I read the article, and suggest you do too. Hmmm.. sounds like a miracle cure, doesn’t it?
After four years supporting the WebTrends web log analysis software in the 1990’s, I’m used to the limitations of log and cookie based analysis, and there are several assumptions in the method described in this article that I would take issue with…
“1. If the same cookie is present on multiple visits, it’s the same person.”
Not true – it’s the same browser / computer, but not necessarily the same person. Take a public library computer: hundreds of people could use that – it’s most likely that on a popular site with no authentication, that a cookie is set once, which would then be used by anyone using that computer.
“4. We know some IP addresses cannot be shared by one person. These are the ones that would require a person to move faster than possible. If we have one IP address in New York, then one in Tokyo 60 minutes later, we know it can?t be the same person because you can?t get from New York to Tokyo in one hour. ”
Not true – IP addresses can be set by a remote data centre / proxy depending on your ISP. The user doesn’t physically have to be in those places, it just depends on how their traffic is routed. The ‘AOL problem’ (many IP addresses designated during any one session) applies here too. Whoever wrote this, doesn’t seem to understand network topography, which I would suggest is fairly useful when talking about Internet user measurements.
Its also states that “30-50% of people delete cookies”…. and then proposes a whole system based on persistent cookies. It’s no good using a range if the range drops by 30-50% in the first place, even if you weight this. Again, a flawed concept.
But I think the most telling ‘wouldn’t touch with a bargepole’ comment in the article would be:
“We’ve approached some household names in metrics, online advertising delivery, and major search engines. The response has been zilch. No one wants to know.”
If online ad agencies and other metrics companies aren’t interested, you must be flogging a dead horse.
To be honest, if you read this without focussing on the techie bits it’s a simple sales pitch for a dodgy scheme.
So the summary answer I gave to the emailed query was “I’ve not heard of this company, I’d not heard of this ‘research’, and I would suggest our time could possibly be better spent.”
On a similar note, an article on the BBC site – Web counting tools ‘need change’ – makes interesting reading, although again, offers no real solutions. I’d be very interested to know if anyone has come across any other methods of measuring users on Web sites.