Have you ever seen a soap opera with twins?  They're always played by the same actor, so they look exactly alike on the surface.  But when you peel back that perfectly coifed veneer, one twin is always good, and the other is always evil (and usually trying to kill the good twin).  If a soap opera can have a moral, twins could represent the lesson "appearances can be deceiving."

Why the digression into soap operas?  Because turnover statistics are the soap opera twins of the business world. 

Imagine you're about to have major surgery and you have a choice of hospitals: AnyTown Hospital or MyTown Medical Center.  Both have new facilities, TV ads that tug at your heartstrings and exactly the same annual turnover rates?16 percent.  They're so outwardly indistinguishable that if this were all the information you had available, you'd have to flip a coin to choose between them. 

But let's pretend that you do a little investigating, and you find that while these hospitals look the same on the outside, they're very different inside.  While they have the same overall turnover, AnyTown Hospital's turnover is heaviest in roles like housekeepers, kitchen staff, and financial office personnel (while very few nurses and pharmacists leave).  But at MyTown Medical Center, turnover is heaviest in nursing and pharmacist roles (while housekeepers and kitchen staff rarely leave).

Here's one more bit of information (and what follows is frighteningly true, not hypothetical).  The Institute of Medicine (a group of the smartest researchers studying the healthcare industry) estimates that up to 98,000 people die in American hospitals every year from preventable medical errors.  Yes, we said preventable medical errors.  These deaths are the result of both system failures and human errors, and can include infections caused by insufficient handwashing, misinterpreted handwritten prescriptions, inconsistent application of known clinical best practices, and more.  The point is that nurses and pharmacists are better able to prevent these errors than housekeepers and kitchen staff.

Have you discovered the evil twin?  Both hospitals look the same (16% turnover), but one of these hospitals is much more likely to kill you (MyTown Medical Center).  Turnover numbers, as they're typically calculated, don't tell you much.  Even if we knew which job roles were suffering the greatest turnover, we'd still be missing information about whether we were losing high performers or low performers.

Imagine you're a stock analyst for a Wall Street research firm and your job is to determine whether certain companies are doing well or about to fail.  You've been researching a large software company (we'll call them Mircospot).  You just got a tip that 400 programmers have abruptly left the company.  Before you can write your analysis about how the loss of these 400 programmers will impact Mircospot, you should probably ask a few questions.  Any ideas?  Here's a question we'd like to ask: Were these 400 programmers high performers or low performers?

If you really want a turnover statistic, try breaking out turnover by high, middle and low performers.  Most executives give lip service to the concept, but most don't truly track this number with the same vigor they apply to their other statistics.  And of course, tracking this number requires that your performance appraisal scores aren't inflated and that the whole process isn't perfunctory and cursory.  Ideally, your turnover number is close to 0% for your high performers and close to 100% for your low performers.  

As the saying goes: There are lies, damn lies, and statistics.  Before you get too fixated on a number, make sure it's telling you what you think it's telling you.