The Cost of Spam

Somebody is way off. The question is: who? This eWeek article cites Nucleus Research's conclusion that spam costs oganizations an average of $1,934 per employee, up by more than 100% from $874 a year ago. This NetworldWorld Fusion article says that KeyCorp calculated that spam cost them more than $1 million. According to their own web site, KeyCorp has 19,585 employees, and multiplying that by last year's Nucleus figure of $874 comes to $1.7 million, so the figures seem to be within striking distance of each other. But, compare those figures to the $10 billion figure that Ferris Research published as the cost of spam to US corporations in 2003 , and do the division ($10,000,000,000 / $874) and you come up with a mere 11.4 million employees. But IBM and Microsoft combined claim about 200 million email seats, and surely more than 5.5% of those seats are in the US and are affected by spam. Somebody is way off.

My own numbers, which are extrapolations based on the Ferris data combined with economic data for the total US economy and some very rough assumptions about what percentage of the US economic output is due to workers whose productivity is likely to be affected by spam aren't in terms of cost per employee. Instead, I've arrived at an average figure of $1900 cost of spam per $1 million of revenue in the total organization. That was with last year's figures, and I want to emphasize that it's just an average based on some assumptions that I haven't thoroughly researched, and I'm reasonably sure that the actual impact on organizations varies from far lower to far higher than that. But it's very hard to reconcile that figure with the Nucleus numbers. Doing so would imply that the average employee is generating nearly half a million dollars in revenue, and that's clearly not true.

So, I wonder... is Ferris's number too small? Or is Nucleus' number too large? The reasonable agreement of the Nucleus and KeyCorp numbers might suggest that Ferris' number is too small, but that's just one data point. I'm not willing to say that's enough evidence to say for sure that Ferris is off, and in fact my gut is that they're probably both wrong but Nucleus is off by a bigger factor. I'd love to see more data points.