Michael Jordan's Weblog

How about counting productivity interruptions?

• Intel has a 20-page document describing how their IT department does a before/after analysis on how long certain tasks take to see if introducing new internal technologies make a difference. (It's not until about page 18 where they actually tell you how they measure anything, though.) I thought the methodology was interesting but it didn't touch on how you tell whether you need a new technology in the first place. This seemed to be trial-and-error -- make a new tool then see if it makes a difference. If not, throw the tool away? It's worth skimming the Intel series (they have several articles on using technology to improve productivity).
• The Boston Consulting Group introduces a term called "Workonomics" in their paper on the topic as a way to prove that measuring employee productivity is important. They observe that common metrics in this area such as sales$/employee are not that useful because they can be skewed and they don't really apply to employees across an organization. (Sales$/employee can be useful at the aggregate level, I suppose, but if it goes in the wrong direction how do you know what to do?) Their paper addresses general measurements related to employees, not really productivity per se.

Measuring employee productivity is tough because you have to make trade-offs any time you choose a numerator and denominator.

I'll offer here that a far simpler method would be to measure detractors in productivity -- technically, I guess we're talking about measuring non-productivity: Defects. A productivity defect could be a system goes down, or it took too long to get to work, or a manager interrupts a developer during coding-time. Whatever!

That's the beauty of the system: the definition of a defect can vary for employee type, and can vary as much or as little as you allow depending on what's important and actionable in your business.

Data collection is a different story, slightly more complicated but not really that big a deal as long as you prioritize the defects you want to track in terms of what defects, if eliminated, would drive down cost or drive up revenue, etc.

Once a set of possible defects is defined, you need to define a period of time (what you are after is a number of defects per a set number of opportunities). For example, if you divide the work day into three periods (say...morning, mid-day, afternoon) then each five-day week has 15 periods or opportunities. Use a more granular breakdown if you want.

The tricky part is then getting employees to track whether any of the pre-determined defects occurred during any of the three periods each day. Read: checklist. This isn't a data collection effort you want every employee performing every day. Consider it a sort of Neilson rating system that you can deployee more or less often depending on your appetite. You want to do it often enough so you can see results.

After each data collection period, aggregate the data, count the defects per opportunity to see which kinds of defects happen the most, and take action. If you take the right action, defects/opportunity should go down.

Leave it to managers to decide what their employees should be doing -- that's a different issue.

Watch the productivity sky-rocket!!

What do you think?

Jeffrey Sonnenfeld, associate dean of Yale University's school of management: "I think Donald Trump is a more positive role model for men than she is for women,'' he told the Mercury News. "At least there is a self-mocking with Trump. She actually believes in the deification.'' (From 2/10/05 Good Morning Silicon Valley)

Yeah, but Carly's got better hair.