Why Websense's FUD doesn't add up

In response to one ZDNet reader's question regarding a previous post (see: Questionable $178B loss: Employees' fault? Or management's?), I decided to, as best as I possibly could, figure out how Websense might have arrived at such a huge number to represent the total annual dollar losses that can be attributed to personal usage of the Web by US employees while on the job.  In its press release, Websense calls this "cyberslacking."  Setting aside any question as to the legitimacy of such a statistic when it comes from a vendor that might benefit from the fear, uncertainty, and doubt (FUD) that stat creates, Websense's announcement does not provide the nitty gritty math behind its calculation. So it isn't pefectly clear how the company arrived at the $178B total.  I've placed a call to the company's manager of public relations Jennifer Culter to get the breakdown.

First, an outline of the data that the press release cites. 

• 68 million US employees have access to the Internet.  The source is IDC.
• 44.7% said web surfing a distraction at work.  The source is an AOL/Salary.com survey with a sample size fo 10,000.  This stat doesn't get used in the calculations (even though the sample is probably more projectable than anything else Websense uses). 
  
• 50 percent of workers surveyed use their  Internet access at work for work and personal reasons. The source is a Websense/Harris Interactive Survey.  No sample size was provided in the release.  Before examining this datapoint, I'll just say for now that, for the rest of this blog, I'll refer to this group as "admitted cyberslackers."
• Use of the Internet for personal reasons while on the job averages out to be 5.9 hours per employee.  The source is the aforementioned Websense/Harris Interactive Survey. 
• Then, the press release cites usage of the average U.S. salary, as reported by the U.S. Bureau of Labor’s National Compensation Survey but doesn't give the figure
  

Via e-mail, Culter filled in several important blanks:

• The average U.S. salary is $17.75 per hour.  I looked for this data on the U.S. Bureau of Labor's Web site and found it, but it's old data from 2003.  There's  a more recent document from 2004 that says it's $17.80 per year with the following qualification: Annual wages have been calculated by multiplying the hourly mean wage by a "year-round, full-time" hours figure of 2,080.  So, had Websense's public relations department been on the ball, they would have used the higher number and it would have tacked on an additional $534 million for a total of $178.534 billion.  But I won't split hairs on this.
  
• The average number of weeks worked per year per full time worker is 50.  No source is cited by I think that's a fair estimate.
• The sample population for the Websense/Harris Interactive survey cited in the press release consisted of 354 IT decision makers and 500 employees.
• The 5.9 hours per week usage of the Internet for personal reasons while at work is an estimate of employees' personal usage by the 354 IT decision makers.
  
• The description of the survey's methdology says "The IT decision-makers survey was conducted online...among a nationwide cross section of 354 IT decision-makers in companies with more than 100 employees" and that "Data are representative of those employees and IT decision-makers surveyed" and then goes on to say "This online sample is not a probability sample."
• The sample was randomly selected as opposed to coming from Websense's customer list.
  

So, using this data, Culter says the calculation went as follows:

• 68 million US employees with Internet access at work  (multiplied by)
• the 50 percent who admitted to cyberslacking = 34 million (multiplied by)
• the 5.9 hours per week that employees are estimated to spend using the Internet while at work = 200,600,000 total estimated cyberslacking hours per week in the US (multiplied by)
  
• The $17.75 hourly cost per employee (not including benefits) = $3,560,650,000 supposedly lost per week in the US as a result of cyberslacking employees (multiplied by)
• 50 weeks worked per year per cyberslacking working = $178.033 billion dollars lost per year due to cyberslacking in the U.S.

So, there's the calculation for you.

But, regardless of the breakdown, the $178B number is still suspect. 

Let's first start with the usage of the 68 million US employees that were used as the starting point for Websense's calcuations.  I checked with IDC to find out if the 68 million employees cited in the study that was quoted by WebSense were full-time employees.  Via e-mail, IDC spokesman Mike Shirer told me "We don't distinguish between full-time and part-time in the survey used to generate this.  We just look for 'business employees' of any stripe."  Therefore, any application of the $17.75 per hour figure (a number for full-time employees) to a number whose composition is not 100 percent full-time employees (eg: the 68 million number used by Websense in its calculations) is a deceptive marriage of two unrelated datapoints.  If the house of cards doesn't fall down based on this point alone, there is more.

Whereas the 50 percent number that was used to estimate the total number of employees who are cyberslacking is based on employees who admitted to using the Internet for personal reasons while at work, the 5.9 hours per week that those employees are implied to be engaged in cyberslacking comes from the IT decision makers and not the supposed cyberslackers themselves.  

Not only that, the 5.9 hours itself is suspect.  By Websense's own admission, the sample population did not come from Websense's customer list and it therefore can presumed that many of them don't have the sort of solutions that Websense makes -- solutions that would have given them some reliable data on which to base their estimates.  Yet, with more than 100 employees for each of the surveyed IT decision makers to be keeping an eye on without such tools, they're still able to estimate the average number of hours that each of the cyberslackers in their companies are cyberslacking.  Hmm.

Another problem.  Forget any of the aforementioned questions I've raised about the number of hours American workers are supposedly cyberslacking.  The $178B total annual losses assumes that all of these hours spent on the Web for personal reasons are indeed hours that the managers of the so-called cyberslackers would view as "cyberslacking."  What if many of the companies surveyed permit such usage of the Web as long as employees are meeting their documented performance goals? In that case, those employees are not cyberslacking and the hours they spend on the Web aren't being viewed in terms of monetary losses by those companies.  Therefore, that "potential loss" cannot be factored into any calculation of total annual losses due to cyberslacking.

There are probably plenty of other possibilities that we could come up with that would blow up the $178B number.  When I questioned the the marriage of the 68 million number to the full time salary data, Websense's Culter responded with statement that deals with any challenges to Websense's assertions.  Said Culter via e-mail:

I think our estimates are in line with the recent survey that Salary.com and AOL put out that "employers spend $759 billion per year" on lost productivity. In that article, 44.7% of employees cited web surfing as their #1 distraction. Also, I think that the "50% of employees that say they surf the internet for personal reasons in the workplace" stat found in our Web@Work survey is a conservative figure—that’s only what employees admit to doing, not necessarily what is actually occurring. For example, in comparison 98% of IT managers surveyed reported that their employees did engage in at least some personal surfing.

All of this may be true.  Websense's estimates could very well be in line with other studies and that 50 percent number may be conservative.  But in its press release, Websense didn't say that these were estimates based on creative accounting that the company thinks are in line with other studies.  In its press release, Websense very definitively says:

Websense, Inc. (NASDAQ: WBSN), the world’s leading provider of employee internet management solutions, today announced that internet misuse in the workplace costs American corporations more than $178 billion annually in lost productivity. This translates into a loss of more than $5,000 per employee per year.

Correctly worded, so as not to confused estimates or opinions with fact, this could have said "today estimated that Internet misuse in the workplace could be costing American corporations as much as $178 billion annually in lost productivity." But the release doesn't say that.