Focus on ‘Reach’, use Rank anomalies, and the ‘geek factor’ for audience, to lower the reach figures to get the real picture for some sites.
Many ‘web 2.0’ startups are likely to have a bunch of their own employees who have the Alexa bar installed and are feverish stats obsessives. For moderate ranked sites this can skew Alexa since their own traffic is a significant percentage of the overall number of Alexa users who hit the site. Fortunatley, you can actually use this to help correct the stats.
Here are some rules of thumb I use to get better stats:
1. Ignore Page views and Rank, for sites that are not in the top 5,000.
2. Always monitor your own site, from a machine that does not have Alexa istalled then look at your site’s reach vs another’s reach and then your sites rank vs another’s rank. If there is a marked difference in the spread between these 2 sets of figures such that rank looks relatively better than reach for the competitor site, then they probably boost their own Alexa figures. Depending on the spread you can actually use this to your advantage to calculate the appropriate reduction in their reach numbers.
3. Weigh in the ‘geek factor’. A site with 1000 geek marketing types will likely have a higher Alexa than 100,000 teens. Look at inbound links to the sites and try and figure out the audience and weight appropriately.