Wednesday, May 03, 2006
Software and Web Development Blog
For sites with recurring traffic there are at least 10 times less unique customers than there are unique IPs.
Example: if an online radio or news site (i.e. a site with recurring traffic serving primarily repeat customers) claims that it has 1 million unique visitors (meaning IPs) in reality they have no more than 100,000 unique visitors.
Explanation: in USA there are about as many DSL customers as there are Cable customers. Presuming that the latter have static IPs the former draw different IP from some pool each time they connect.
Since recurring traffic presumes visiting the same site over and over again, quite frequently from different computers (i.e. at home, at work, via laptop, etc.) new IPs are drawn from the pool disproportionately inflating DSL customer visits.
While it is obvious that DSL visits would be inflated they are drawn from a limited pool. Given enough time the entire pool will be drawn by DSL customers thus giving the impression that there are as many unique visitors as there are IPs in the pool.
So unless a site is really big (like CNN.com or MSN.com) the number of actual unique visitors is much less than the size of the pool, hence the number of unique visits will be inflated by the factor of times each customer visits the site in any given interval of time.
Hence if the number of unique IPs is measured within a month and it is known that via random callout that visitors visit the site at least 10 times a month then the equation linking unique visitors with unique IPs will be:
IPs = Visitors*(DSL_FRACTION*VISITS_PER_TIME_PERIOD + CABLE_FRACTION)
If DSL_FRACTION = CABLE_FRACTION = 0.5 then we get
IPs = 5.5*Visitors or Visitors = IPs/5.5
And the most important unknown factor is the number of visits per time period.
So do not trust stats, they ARE inflated.
source:http://techsearch.cmp.com/blog/archives/2006/03/online_advertis_1.html?loc=software_and_web_development
