Weird little piece on Upmystreet in this week's Guardian Online section

The piece carries some stats that just don't add up for me.

These days it's only the poor old Government that whitters on about hits rather than users. Pity them as they seek to justify the money they burn on tripe.

However, when it comes to claimed users, you can get your log-derived statistics audited, like Upmystreet did in January. They claim 660,000, but they also paid ABC to verify that claim. Newspapermen will be familiar with ABC - even before this new-fangled interweb came along, they did the same job so that advertisers could trust publishers and vice versa.

But how do you estimate for a non-audited site? There's a few tricks that every hack should have in his repertoire:

• Alexa rankings don't tell you an absolute figure, but they do allow you to estimate traffic relative to other sites. Furthermore, it's very likely that the distribution of traffic across sites follows a power law, so those below 70,000-odd probably receive several orders of magnitude less traffic than those in the top 7,000.
• Links known about by Google. While it's theoretically possible that a site could get all it's traffic from keen typists, it's very very unlikely, so a site that doesn't have much google knowledge, doesn't have many visitors.
• If they're a community site, you want to look at their messageboards. Are they busy? Even the crappiest of community sites must have a read-to-post ratio of about 1%, so you can get yourself an order-of-magnitude figure for visitors by multiplying visible posts by 100. Shoreditch in March

• If it's a network of sites, then obviously the picture is more complicated, but if even their flagship site languish in the doldrums you might decide to get suspicious. Especially when you can compare it with sites whose traffic you know for sure.
  

If, as a hack, you suspect someone might be fibbing to you about their site traffic, you may have to consider whether they would fib about other things....

NB Interesting little side project for someone to build: take audited traffic numbers, and plot them against alexa rankings. From the curve (if it was good), you could then estimate traffic numbers for all sites.