How Unique Is A Unique Visitor? | Hacker News Reader

How Unique Is A Unique Visitor?

avc.com

14 pointsalexandros8 comments

And what they found was that a "reliable cookie" overstates user counts 2 to 4 times. That's right, if your analytics software uses cookies, it is possible that your unique visitor counts are 2 to 4 times too high.

I wonder what their methods for determining this is. Do they mean that an average user clears their cookies 2 to 4 times within a typical reporting period (unlikely)? Do they mean that an average user uses 2 to 4 different browsers to access the same site (unlikely)? Or do they mean that the average user accesses the same websites from 2 to 4 different computers (more likely)?

These kind of claims don't mean anything if you don't know how to interpret it. People know how to interpret unique visitors: it essentially means "unique browser sessions", and most people that are serious about analytics are aware of this. But don't start dividing the unique visitors by 2 to 4 yet, if you don't know what the resulting number exactly represents.

(disclaimer: I'm a service provider in online ads and biased too)

Correct me as needed, but I took OP to regard "unique visitor" as meaning "unique visiting person" - and to be citing himself as overcounted when he visited a site variously via Safari, FF, Chrome, and iPhone - which certainly happens to me many days.

I'm not sure those averages are so unlikely. Sure, the "typical" (median, 75th percentile, whatever) user probably doesn't clear cookies or switch browsers that often. But it only takes a small number of users with unusual behavior to raise the average by a lot.

If just 1% of users have cookies set to expire on exit, or use a "private browsing" mode, or whatever, and those users visit the site an average of 101 times per reporting period, then you have a 2x overcount already.

101 visits per month is probably high for most sites, but if you also have a significant minority of users who visit your site from two or more computers, that will raise the average too.

Tracking uniques with Flash cookies rather than browser cookies (when possible) should compensate for the one-user-many-browsers problem. I know there are ways for individual users to clear Flash cookies, but I don't believe this is being done on a scale wide enough to be statistically significant.

That sounds a lot like trying to solve a small problem by creating an even larger problem.

Translation: whatever the problem, it's unlikely the solution is using more Flash on your website.

Half the "unique hits" on my old Blogger are me.

It sounds like some javascript was used to hash some things like installed plugins that vary from user to user and browser install to browser install in order to create a more unique and reliable identifier than cookies. I'm not sure how that can capture use of multiple browsers though.

If they are on the same machine you can just use Flash cookies since there is a single Flash cookie jar for the entire machine (across browsers).

Reply on news.ycombinator.com