Ian Welsh

I want to talk a bit about management measurement. I recently spent a number of years in a good sized multinational, and I watched management trying to gain control through measurement. And mostly I watched as they gained the wrong sort of control; as they crystallized behaviour in ways that lose more from employees than they gained.

(This is an old, old piece, one of the few I saved from BOPnews. Originally written in 2004 back when I was still corporate. Since almost no one will have read it, and those who do won’t remember it, here it is again. I’m putting it back up because it relates fairly closely to the recent article on the lack of belief in good and why incentives rarely work.)

When you’re dealing with small numbers of people, simple measurements are all you need, and indeed the time spent measuring can be a simple waste of time. For larger groups, and as management becomes disassociated from the actual work of the organization, measuring is necessary so that management knows what is happening and can modify it. The old saying (which I’m sick of) is that “you can’t manage what you can’t measure.” It’s a statement with a lot of truth to it, but so is this – “you measure what you manage, so you’d better be sure you’re measuring what you want to manage.”

Here’s an example. A friend of mine used to do customer support for laptops. He was measured on how long he was on the phone and how quickly he picked up. If he spent too long on the phone on average, then he was taken aside and reprimanded. These measurements encouraged tech support employees to get people off the phone as quickly as possible, whether their problem was solved or not. Assuming management actually wanted happy customers (ie, that they saw tech support as a way to sell the next laptop, rather than something they had to do as cheaply as possible) then the way to measure this would be to have an automatic survey at the end of the phone call, asking how satisfied the client is. Since there will always be jerks who are never happy with phone support, you set the threshold at a certain percentage of “unhappy” customers and then if someone goes over that you investigate. To keep productivity up you measure phone time and compare to satisfaction ratios and (horrors) investigate individual reps who spend more time than normal on the phone, then coach them individually on how to solve problems with less chit-chat while still keeping the customer happy.

I’m going to discuss five issues related to measurement. The first is the problem of measuring what you can easily measure. Simply put, it may be more difficult to measure some things than others. Management tends to measure those things that are easy to measure. In a call center there are plenty of systems which will allow you to track a wild variety of phone stats, but you can’t measure one CSR helping another with a call. In sales you can measure how many sales a salesman makes and how much they’re worth, but it’s more difficult to measure whether he’s made verbal promises your company will have trouble living up to. You can measure the number of code lines a programmer put out, but it’s harder to measure how easy they will be to maintain down the line.

This is often a systems issue. Whatever the system assists your employees to do, is easy to measure. So if you have a system that presents work items, and which employees close those work items, it’s easy to measure how fast they’re doing them. But what if some work items are harder than others? And what happens to those employees who are taking calls or e-mails you can’t track and are helping customers or other employees with those problems – is that behaviour you don’t want to encourage? Because if you’re measuring only processing times then those who do other things will be measured as less productive. So they stop helping customers, and soon you have a reputation as having unresponsive employees who never want to take time to help people.

And this leads to the second issue, which is what I call Putting your Fingers Down. Another way of putting it, is “you get the behaviour you measure.” If a job involves 10 activities, and you publicly measure only 5 of them, your employees will gravitate towards those activities. It often seems obvious what an employee does. Let’s say you have repair techs in a retail store and you decide to measure their productivity by measuring how many appliances they repair. Sounds good eh? Productivity increases and you’re happy.

Until you start getting complaints that the repair techs don’t want to talk to customers, and that when they do all they seem to want to do is get away from them. You also hear that some techs are taking easy repairs and leaving the hard repairs for others, who put them off, because that boosts their stats. So easy repairs are getting done fast, the hard ones are getting done slower, and customers aren’t getting individual personal attention any more, so they aren’t happy. That worked well!

Which leads to what I call the The Limits of Coercion. Public measurement is a form of coercion. The idea is to measure people and then push them to do better and get rid of the ones who don’t measure up. You put your fingers down and say, “do this!” And you can absolutely do it. Whatever behaviour you are able and willing to take the time to measure, you can and will get. But what you can’t get is positive cooperation. You can’t make people do the extra things. And people resent the wrong type of measurement. The problem is that you as management think you understand the job. Problem is, unless you still do it yourself, you probably don’t. Outside of the sort of jobs that are truly subject to Taylorization, most jobs require a myriad of little tasks and if people don’t do them, the overall job suffers. If you start measuring the wrong specific things then people’s attitude when you pull them in for a talk is “I’m doing fine on the stats you said you want, I don’t have time for the other stuff.”

The other problem is that people subvert the measurements. There are almost always ways to make the numbers come out better than they should, and people will take the time to find them and do them. Which leads to the fourth issue, the question of “Public metrics and private metrics.” Simply put, when you’re setting up metrics you should first find out which metrics track each other; figure out why they track each other; and measure both sets. But one set you keep private and the other is the public set. If the private set starts diverging from the public set then you should investigate if people are fiddling with the public set. Odds are they are.

But the real, final point is that you should be looking for your “Bottom Line Metrics”. In a call center it might be the percentage of happy callers divided by the average time per call. In a processing center I once worked in the VP (a very wise man) used to publicly (I’m sure privately he had a number of measurements which had to remain satisficed) measure only one thing – the average time from a piece of work entering the center to the time it left. He didn’t measure any specific processing times – only how well the center was working overall. If that number went up he’d want to know why, and when he wanted it to go down he let people tell him how they were going to get it down, not the other way around. The center ran very well. When he left his successor started putting his fingers down and both customer satisfaction and employee happiness declined.

In the end you should ask yourself “what are we trying to accomplish?” Then you publicly measure that, and only that. It may seem that you want to do multiple things, but in most cases you can boil it down to one thing – as with the customer service center where happiness was divided by call times. You want people to go away happy after their call with the least time necessary to make them happy. If you can’t break it down then you either don’t understand what the job actually entails (or what your division or company does) or you may need to break the work into different functional groups.

Finally, don’t fall into the MBA trap. As a manager you probably don’t really know what your employees are doing. You probably don’t really understand what is required to do the job well. However unless you’ve beaten them down too hard, or you’ve got a crew of reprobates, most people want to do a good job. Most people want to be able to say “damn, we’re good!” Don’t treat them like untrustworthy children, and you may find that they’re on your side and that measuring only the bottom line, on the minimum, is sufficient. When you go to war with your employees and try and measure every specific behaviour, generally both sides lose.

(Originally published in 2004 at BopNews. Republished April 17, 2009.)

(My writing helps pay my rent and buys me food. So please consider subscribing or donating if you like my writing.)