Ricochet is the best place on the internet to discuss the issues of the day, either through commenting on posts or writing your own for our active and dynamic community in a fully moderated environment. In addition, the Ricochet Audio Network offers over 50 original podcasts with new episodes released every day.
The Pitfalls of Management by Measurement
The news is full of Wells Fargo’s follies since they got hit with fines totaling $185 million last week. What happened was a case of management by measurement. Wells Fargo employees were heavily pressured – including threats of job loss – to rack up customer “solutions,” which translate into selling additional services to the bank’s customers that included additional bank accounts and bank credit cards. In order to meet the strict quotas that management had imposed, employees opened accounts for customers without first receiving customers’ permission or informing them.
Though this practice was widespread (some 5,300 Wells Fargo employees have been fired since 2011 for opening fake accounts), it does not appear to have been the result of a conspiracy. Rather, it was an example of spontaneous order that emerged from employees acting in their own best interests — in this case, reducing the pain of management pressure — given the incentives and constraints imposed by the system. In an article appearing on Bloomberg, Matt Levine explains how this sort of thing happens:
Two basic principles of management, and regulation, and life, are:
- You get what you measure.
- The thing that you measure will get gamed.
Really that’s just one principle: You get what you measure, but only exactly what you measure. There’s no guarantee that you’ll get the more general good thing that you thought you were approximately measuring. If you want hard workers and measure hours worked, you’ll get a lot of workers surfing the internet until midnight. If you want low banking bonuses and measure bonus-to-base-salary ratios, you’ll get high base salaries. Measurement is sort of an evil genie: It grants your wishes, but it takes them just a bit too literally.
I’ve run into the same sorts of unintended consequences of the “if you can’t measure it, you can’t manage it” mindset. For example, getting support from our IT department required that a “ticket” be opened describing the issue or problem. Management decided that ticket count was a good way to measure productivity, so IT supervisors put pressure on analysts to boost ticket numbers. One result was that tasks that used to require a single ticket were suddenly divided into multiple “sub-tasks,” each requiring its own separate ticket. As a result, IT work entailed a lot of unnecessary overhead. It got so bad that you needed a ticket just to talk to your database administrator, whose office was a couple of doors down the hall.
Do you have any examples of “mismanagement by measurement”?
Published in Culture
Funny you post this. I experienced that Wells Fargo experience first hand…as an assistant brand manager in the late 90’s, cross selling was the tool, and submitting credit card applications without express approval was the norm. The compliance side of the business put a stop to it locally whenever they directly came across it, but the sales management side turned a purposeful blind eye. I couldn’t get out of that job fast enough. But thats sales for you. Blehhh!
Regarding your IT note…the problem isnt with tickets and measurements, its how its utilized. Unfortunately, most team managers are former practioners, and aren’t necessarily savvy enough, or predisposed enough, to learn how to effectively manage via the numbers, without coming across looking and feeling like its the numbers that are motivating them.
In my first job, the building had few windows (other than “Mahogany Row”), but one was above the entrance where most employees entered the building. One of the managers would stand at the second floor recording which employees came in after the nominal start time of 9:00. I was a software engineer and this was in the era when the computer was not time shared, but you signed up for a block of time. This was before wife and kids, so I often had the late night block – from 10:00 or so until around 2:00 AM.
I finally got upset at the measurement and told him that if he wanted to record my coming in, he should record when I left – yes, I was that sort of guy in my youth.
I have always felt that measuring any creative work by a timesheet was the epitome of dumb management.
Over 25 yrs now in Manufacturing and Supply Chain, I’ve seen this come up endlessly. I’ve gamed measurements to alleviate management pressure. I’ve built tools to catch others gaming measurements. I’ve worked in multiple settings where management can’t seem to stop creating more metrics to follow. And can’t understand how various metrics can sometimes work against each other.
My biggest complaint though is not that too many things are being measured (though often that is the case) but that little to no time is spent analyzing the results for root cause, and fixing problems. Data rich, analysis poor. The group I’ve just joined in my company has recently been tasked with developing more metrics. Sad.
Third year in a row now of a major push to reduce inventory, and third year in a row of plants being forced to take extra days down to help reduce inventory (even plants with sizable order book backlog). Of course, sales gets pressure to show sales growth. So first half forecasts tend to get a little beefy, and plants are running based on forecast signals. Lather, rinse, repeat.
I doubt this problem is related to just Well Fargo, but is most likely a standard financial practice.
I know a few people working low level retail sales. I am amazed at the pressure they are being put under to get people to open credit cards / lines. In many cases there are monthly quotas to open new account and if not met the employees are written up and eventually fired. One funny story a friend that works in the mall tells me is that most of the accounts they open are for other mall employees from different stores and not mall customers. It seems that most stores have these quotas so in order for everybody to meet their numbers the mall employees will fill out these applications for each other so as to help each other out and keep their jobs. Sounds stupid to me but welcome to corporate America where everybody has an angle.
Do you have any examples of “mismanagement by measurement”?
See Stupidity–Communist-style and Capitalist-style
A manufacturing company found that in their effort to reduce workplace accidents by rewarding the team with the least accidents, the only thing that was reduced was the reporting of the accidents.
A customer, whose implementation project I was managing, decided to reward the departments that used the least office supplies, hoping to encourage less waste. The department that “won” was the one whose manager somehow found out before the announcement and stocked up on supplies before the program started.
Another I was told about was the executive who read that low absentee rates indicated high morale. So the word came down that sick leave would be frowned upon. So sick people came to work and infected others…
Here’s my theory of management: Hire good people, give them clear and reasonable goals, then get the hell out of their way.
I read that Bloomberg column and had one instant reaction: managers who set the unrealistic goals probably weren’t included in the firings. Tim
I was one of those. I took a job at Sears while I was underemployed, and being older than their usual part timers I understood that a credit card is not something someone signs up for because some kid offered them a one time 15% discount. At one point I ended up giving a credit card to a Hispanic man who barely spoke English, and I’m still not sure he even knew what he was signing up for. Four years later I still feel guilty for that.
Thankfully I was only there for two weeks before I got my current, full time job.
…Which probably did wonders for morale…
Management make mistakes and correct them and if they don’t, they suffer losses and ultimately bankruptcy. Big data is prone to big mistakes and in large organizations acquires the momentum of expertise and the illusion of certitude big data creates , but still unless they are monopolies (which means enjoy some government protection) they fix themselves. Government uses big data as well, but there is no corrective mechanism so the momentum of expertise just rolls on and over us. The reason the private sector works and the government doesn’t is not that the private sector has better more clever people, which on average they simply do not. It’s because there is constant feedback and daily adjustments to everything. In government none of the feedback and daily adjustment occurs.
Speaking as a CEO:
If people would just do their jobs as they agreed to when they were hired, we would not need to measure anything. This all stems from the fact that people don’t want to work all that hard at their job because they are not engaged.
According to Gallup, less than 10% of workers, worldwide are engaged. In America is jumps to 30%. While you need to measure people, frankly, I find it less about production and more about being able to get rid of people without a lawsuit, even in a Right to Work state.
My goal, and this is going to take time for me, is to get the culture changed to get employees more engaged in their jobs. Wish me luck.
Excellent points.
As a claims adjuster, we got measured on closures and time to closing. It is certainly an important measurement. You need to know whether your adjusters are closing claims. It matters for customers and for management. But it could also lead to perverse outcomes– like closing claims quickly and working them as supplements to keep your numbers up. But this is disruptive to the customer because they may not understand the supplement process or if the timing gets awkward, they may get sent to a new adjuster and feel they have to start the process over again or become worried that agreements they made with an adjuster (informally) may not be honored. This becomes even worse in theft claims where sometimes you need time to properly investigate the claim. You may be tempted to pay on some claims with fraud because you don’t want so many claims to sit in your pending.
But it is also bad, even without the incentives, because if you are in some dispersed locations you just can’t close that fast. Sometimes you are an hour and a half between claims each day for months. You can’t close 3 a day that way. This penalizes the adjuster for doing their job. So people would take a long time off in hopes of getting transferred.
Counting field supplements was also controversial because it created some perverse incentives as well. It takes good managers to see through the numbers.
It would seem that if you have an outlier in the data its the job of the supervisor to explain this problem.
I was talking about metrics with another manager and he told me that there was a lady on a former team he was leading whose average case length was longer than everybody else’s. It turned out that she was taking on all the hard cases of a particular kind that the other techs weren’t as good at and these cases just took longer.
Upon finding this out it ceased to be a problem. I think the difference between good management and bad management is what action does the data prompt.
I had an employer who sold coffee to employees with vending machines. The price started at $0.75 and went down a nickle every month without a lost time accident. Upon an accident the cost went back to $0.75. The result was the price dropped down to free. When someone got hurt, they toughed it out. No on want to be the reason all their coworkers would have to start paying for coffee.
Coffee is for closers.
Every system can be played – and will be, sooner or later. I think of the VW/Diesel story as a perfect example of how engineers solve for the test. But everyone does it, whether it is a KPI system or measuring hours worked per day, etc.
Even CEOs do it – they work to maximize their returns (corrected for risk mitigation) based on their own compensation formula.
None of this is remotely surprising. It is basically impossible to get a normal employee to prioritize the organization over their own limited career goals. It is a key reason why the people in my company are shareholders (owners) first. People who think like owners are very differently incentivized than those who punch a clock, and they act accordingly.
But for better or worse, the vast majority of people cannot actually wrap their heads around what it means to think like an owner. They have been “sticking it to The Man” for too long.
Getting workable performance metrics is a bugbear, especially for things like healthcare. There is no objectively factual way to measure results for cancer treatment, for example, because one cannot control for all the unique circumstances for each patient, treatment regimen, ways of measuring success, etc. Imagine, for example, a metric that uses “survival” as the goal, versus one that uses physical or mental functionality. Very different treatments will be selected based on the metric that has been handed down from On High.
Another reason why the customer (not a bureaucracy) should make health decisions, with advice from their doctor.
Enron. (From a distance and as described to me.) While their financial machinations were first class (though corrupt), their basic project development was haywire – they paid development bonuses based on board approved project pro formas – at financial close. Projects didn’t even have to operate and the developers were hauling in lucrative bonuses. Needless to say, they got lots of great looking projects with wonderful financial models. A project would also get rated on how well it had laid off various types of risk, even if ALL of the counterparties were Enron affiliates. (I had to go look for my eyeballs on the floor after I heard that one.)
A different outcome. I once worked for a company that had in house physicians in their on campus health center (reduces employee down time for doctor visits if they use them on campus and improves health if going to the doctor is easier). One doctor was an outlier and it was easy to see why. He was Mayo trained and he insisted on data. If you were drawing blood anyway, why not take a little more and test for everything that might be relevant. When you went in for your physical he would take as long as it took to go over your lab results with you, the implications, the health concerns and the life choices that might make differences. An office visit often took over an hour. He was always running behind and eventually took over control of his own calendar so he could book the time he thought he needed. It drove the bean counters nuts.
The rumor was that he was going to be fired – miserably failed the metric of patients seen during the day. But the execs wouldn’t hear of it – too many felt he had saved their life by solving health concerns or identifying issues, prior to their being life-threatening, but that needed to be addressed and would have eventually become life threatening. So they hired another doctor and Doc A agreed that he would see his existing patients, difficult cases and the executives.
(to be continued)
(continued)
I am one of those who credit him with saving their life, as his obsession with data led him to want to establish PSA baselines long before recommended PSA testing, which led to his discovery of my early onset, fast developing prostate cancer. Next month is 10 years cancer free.
Scott Adams has a story about a company that paid it’s software developers a cash bonus for fixing software bugs – including in code they had written themselves. The cartoon based on the story concludes with Wally walking away after being told about this, saying “I’m going to go write myself a new minivan”.
I witnessed an interesting corollary to this measurement debacle where I worked for 18 years in a growing company. When management subscribed entirely to the measurement tool, then personal responsibility and accountability plummeted. If an employee did something stupid, lazy, destructive or otherwise injurious to the bottom line, then the first act of management was to examine the process under which the malfeasance occurred and determine which metric was at fault. It had to be the metric, had to be. Those of us who worked assiduously for customer satisfaction were dismayed to see this. And the idiot was never fired, and at most would be reassigned, just like a crappy schoolteacher often is.
What’s most surprising to me about this is not that it happened, but that something of this scale (5,000 employees!) went on for so long undetected. It speaks to very poor monitoring and compliance processes and raises bigger concerns about the company.
And it comes up a little short. Effective managers will follow-up to insure that the thing being measured is authentic, otherwise management performance is superficial.
I worked in rental cars. Branches were graded according to the percentage of their cars were on rent at 2 PM. So of course, every rental possible was closed at 2:02 and confirmed rentals for later in the day all began at 1:58 (charges starting when the rental actually began).
Customer satisfaction phone surveys were of the utmost importance. Therefore, customers we knew weren’t going to be happy (whether because of genuine complaints or just because they were an idiot) were told we ran out of cars and couldn’t rent. Or, after they got the rental, a fake phone number was put into the system for them before the rental was closed. Management got wise to this and set it up so they got an alert every time a phone number was changed. To get around this, local branches just put in the fake number at the beginning and put the real number on a memo line.
And the amount of time we spent ensuring that money went into the revenue category that was pushed at any given month was absurd. When there was a push for more fuel charges, instead of increasing the rate for a better car, charging a late fee, charging detailing fees, or basically anything, we’d charge it all as fuel. Until we had enough in our fuel line item, management would pick another goal, and we’d put everything that was going into fuel into that instead.
Another way to think about this is that you have to know the data. What are your goals? What affects these goals? How do we measure those things? What things affect what we’re measuring? Are those things random & independent or must we correct for them? If we must correct for them, how do we do it? If you’re thoughtful about metrics you can anticipate most of the problems up front. Unfortunately, this can require considerable effort and it undermines confidence in that “dashboard” view of the business that’s so fashionable right now.
I see the good management example you provided as identifying the problems with the metrics after the fact. It’s certainly better than making bad choices because of flawed measurements. It’s also necessary because no a priori system of performance indicators is going to work entirely as intended. But there needs to be a feedback loop that fixes the flawed metrics. If the answer is to trust managers’ instincts regarding which metrics are important and which outliers to ignore, I’m not sure why we wouldn’t skip the metrics and just trust the managers.
Where’s the problem? It sounds to me like the employer’s plan worked just fine for the employer.
This is obviously true when you take one or two measurements and tie them directly to incentives. It’s probably also true if your metrics are more complex and tied directly to incentives. But you’re probably safe if you have a complex system of metrics not tied directly to incentives. It requires a lot of effort to game such a system and offers little to no reward.
The value of metrics is in monitoring trends, assessing the effects of changing conditions, and informing decisions. If a few KPI’s can be used to reliably make all the important business decisions, then executives are superfluous.
When it comes to incentives, every system known to man sucks. All you can do is use the one that sucks least for your business at any given time.