In medicine, an iatrogenic disease is one that is brought on by a medical treatment itself. An example would be when a physician treating a minor condition fails to properly wash his hands and as a result gives the patient an infection more serious than the original problem.
It strikes me that iatrogeny also occurs in the management reporting and control systems of businesses and other types of organizations. A particularly awful example was reported in Britain a couple of years ago: hospitals were being measured on time from a patient’s entry into the emergency room until the time that patient was seen by a physician. It appears that in quite a few cases, the optimization of that measurement for the hospital was achieved by leaving the patient in the ambulance, in some cases for as much as five hours, so that the clock on the measurement would not start until the criterion was certain to be achieved.
So a measurement intended to improve patient service had the opposite effect. It directly caused unnecessary pain and danger to the individual ER patient who was kept in the ambulance while harming the effective utilization of expensive vehicles and skilled personnel, while at the same time providing upper management with a distorted picture of what was really going on.
Smirk not, fellow capitalists. While this particular example of iatrogeny was perpetrated by a government entity, plenty of examples can also be found in the private sector. Indeed, I saw an interesting example in a Target store just the other day.
Targets have customers service telephones scattered throughout the store. If you need help in finding something, you pick up a phone. A voice response system says “Welcome to Target” and then directs you to press one key if you want to talk to a store operator and a different key if you want someone to meet you where you are.
I question the overall design of this system–very few people like interacting with voice-response systems, and having to deal with one of these things when you’re already in the store is probably mildly irritating to many. But the really interesting part of Target’s internal customer service system begins when you select the “meet me here” option. Fairly quickly, an employee does come trotting over–and his immediate concern is to push the reset button on the phone. Obviously, there is some sort of timer which is measuring response times, and a “response” is assumed when the reset button is pushed. The employee is usually in such a hurry to push the button that he does it before even greeting the customer, much less asking the customer what he needs. Sometimes I have observed an employee almost run over the customer in his hurry to get to the button.
On Saturday, the employee came quickly after I requested help. He pushed the reset button, and the system did not reset. He then pushed it several more times. Still, nothing happened, and the timer was apparently in the “no response yet–still running” mode. He then did the right thing, turning away from the telephone and asking me what I needed. Almost immediately, his walkie-talkie started squawking, with a woman demanding to know why the call hadn’t been answered and then telling him in no uncertain terms to get the thing reset. She also apparently told him some alternate way of getting the system reset, which finally was done.
The message this scenario sends to customers–and employees–is that Target cares far more about optimizing its internal measurement of customer service than it cares about optimizing actual customer service. And this is not a problem with individual employees or with individual store managers. Almost certainly, the correct diagnosis would be serious cluelessness on the part of senior executive management.
The harm done by Target’s iatrogenic measurement system is of course trivial compared with the harm done by the bad measurements applied to British emergency rooms. But both are symptomatic of the same kind of problem, and the aggregate harm done by measurement and incentive systems which misdirect is surely very substantial.
Designing effective measurement, incentive, and control systems is hard work. You have to really think about what kind of human behavior the system is likely to encourage, and then watch it and see what it actually does–and make changes when necessary. Too often, these systems are put in place and feedback from local management and front-line employees is ignored or even discouraged.
What other examples of iatrogeny in management control systems have you encountered or heard about?
A couple of related stories…
Another Target story
and one from Wal-Mart
Some years ago the company I worked for was using inventory levels of raw materials as one measurement that went into the goals that the bonus of managers was based on. One manager had a ship full of a petrochemical sit at anchor for 2 days before off-loading the material, thereby putting it into his inventory after the bonus calculation period ended. The demurrage charges were somewhere in the neighborhood of $1,000 per hour, if I remember correctly, but they were charged to some other department.
Having worked a lot on productivity improvements, I think it is very easy to underestimate how hard it is to develop good metrics. Ideally, a metric needs to be easy to collect data on, simple to understand, connect meaningfully to a customer requirement and be stable enough that you can identify a meaningful change quickly. Unfortunately, because the first two requirements are so critical (trying to assess customer satisfaction with a 20 question survey after each interaction only works for high value transactions like car sales) it is easy to end up like the drunk looking for his keys under the streetlight after losing them in the dark alley.
The sad thing is, when Target first rolled out the system you describe it probably worked – it lets the staff know it is important to management that they get to the customer quickly and they’ll be held accountable immediately, which is critical to changing behavior. But once the initial change is made, further improvements are a lot harder, so management starts thinking that while they can’t control how often more customers are asking for help than there are people on the floor to help them, they certainly can make sure their people press the button ASAP. And so what was a tool to drive a desired behavior turns into a counter-productive exercise in button-pushing.
Customer service metrics are a lot like variable compensation metrics. To get what you want you have to tweak them constantly to prevent the organziation from confusing the metric with the actual needs of the business.
We did have a case once where a commercial manager shut his plants down for a few days before the end of a quarter to hit an inventory target. Unfortunately, this just backed up the supply chain to the point where another plant, which was running flat out to build inventory for the upcoming peak season, ran out of storage and had to shut down. And four weeks later we had to go on the market and buy the material the plant would have made at a premium. Total cost was well in excess of a million dollars, and there were some very frank discussions to the effect that no one was going to make that particular decision again without upstream supply chain approval. Although to be perfectly honest, I put most of the blame on the his supply chain folks, who really should have been able to convince the manager this was a bad idea. One of the functions of a supply chain manager is to warn commercial people what it will cost to make certain decisions.
How about the government metric that wind and solar electricity HAS to be taken, even if it means spilling water at a hydroelectric plant? This happened at the Bonneville Power Authority on the Columbia River. The decision that wind and solar had priority was endorsed by a Federal court too.
Dumb, dumb, dumb.
I spent a year and a lot of money at Dartmouth in 1994-95 getting a masters in quality assessment in health care. I forget what the actual degree said but I was interested in how you measure quality in medicine. I had been on the board of the Medicare peer review organization in California for a few years and I had just retired after a 14 hour back surgery. I thought a potential second career was available, as well. I was wrong about the career as I found no one, absolutely no one, was interested in quality. It was all about cost, which can be affected materially by quality improvement, but nobody in an expensive suit believes that.
I also learned a lot about academics in medicine. Jack Wennberg, who ran the program, is a very bright guy and pretty open minded. He has written a lot about variation in health care. If you find a lot of variation in the frequency of something like back surgery or tonsillectomy, the chances are that there is no consensus about the necessity of the procedure. That is why tonsillectomy has disappeared from medicine for 30 years.
The problem is that many of his colleagues, like Donald Berwick, are ideologues and committed to single payer with rationing by a committee who know better than you what is good for you. I got tangled up a bit with one of the young faculty, who writes a lot of BS on health policy, because I favored medical IRAs. I did a paper on them with a ton of references that he didn’t like at all. He didn’t care about previous studies that supported the idea.
That fall, the Republicans took Congress and I got excited at the thought that maybe I could get a minor staff job and help with policy. With the help of Senator Gregg, a nice guy, I got some interviews. I found out that tax lawyers write all GOP sponsored health care bills and they were not interested in input from doctors, even those with experience and a degree from Dartmouth. A couple of sympathetic lobbyists told me there was no chance. I gave up.
I loved Dartmouth, the place, and they even offered me a spot if I stayed. Over the next two years, I looked for somebody who was interested in measuring quality. No luck. It’s too bad because I think you can do it. A few years later, I got hooked up with an outfit that was into the Toyota system of quality improvement. They were very enthusiastic and had a track record in industry, specifically the aircraft industry, and thought health care would be a good prospect for growth.
They had no more luck than I did and the whole thing went away. Health care is tough but the biggest barrier is the admin people. All they care about is numbers, hours worked and productivity. The only thing that stands between us and nasty bureaucrats is the private system we have. Patients have choice so far. Even the HMOs have to worry about comparisons to private fee-for-service medicine. Canada had a comparison to the US private system and private care is returning to Canada. The French, who have a fee-for-service system that is excellent, can’t get British immigrants to go home to the NHS for care. This is breaking the bank of the French system.
If we ever get single payer, or even real Obamacare, it will be managed by numbers by the people who are running the Target phone system. Massachusetts is already trying to make participation in the state health plan mandatory for license renewal. If we don’t watch it closely, we’ll get another example of the British ambulance story. I’m not a believer in the malice of insurance companies but they are mostly ignorant about health care. It’s all about numbers. Hospital administrators are probably worse than insurance executives.
See also stupidity, Communist-style and Capitalist-style
I think a better approach to the Target situation would have been: first, to use a device other than a phone. For example: a wireless device with a speaker/microphone and 2 big red buttons, one for “meet me mere” and another for just asking a question. For the timing of the response, have the employee carry something in his pocket that automatically activates the reset via proximity—maybe RFID devices would work for this. There are dozens of electronics firms that would be capable of designing and manufacturing such a system in appropriate quantities for Target store use, and the psychology of interacting with such a device would be much nicer for the customers than a voice response system. (The typical I/T person, though, is unlikely to imagine such a solution, being more likely to think in terms of hardware & software available from his known vendors.)
More important than any technical solution, though, is the cultural aspect. I don’t know the details of how Target evaluates its store managers, but if excessive attention is being given to top-down-directed operational metrics, and too little to revenue/profit/customer retention, then misdirection will occur.
As Phwest said, the behavioral implications of metrics are very tricky, and the effects need to be carefully observed and the metrics modified where necessary.
A study recently reported in HBR showed that–unsurprisingly–those companies in which employees identify with the company tend to do better at selling stuff, and furthermore than *management* identification with the company tends to improve *employee* identification with the company. (Insert disclaimer about correlation and causation.) When it becomes apparent to all in a company that the metrics in use are silly and even counterproductive, then employee identification and management identification will surely both suffer.
I have read several of Suvorov’s books. I believe he was the defector who chose the British instead of the US because, as he said at the time, the Americans would screw up his defection.
One other story….Peter Drucker said he knew a railroad where customers at small rural stations were always walking off with the restroom key. Getting a new key made was a capital appropriation and required a 3-month approval process; whereas “emergency repairs” were within the station manager’s discretion…and every station had an axe…
Not sure I believe this (capex appropriate for a key??) but it makes a great story.
A bit over a decade and a half ago, all state employees in Colorado were placed on something called “Pay for Performance”. The former 7 pay steps you went through your entire career [5% per step over a 30 year career] were voided, so you get no raises from hire to retirement except for the rare COLA unless you are promoted. Instead, the legislature promised to appropriate half the money used for step raises for P4P one-time bonuses based on annual evaluations for those with an “Outstanding” rating. In point of fact, they rarely appropriate any money for it and never have fully funded it.
In the Department of Corrections, the evaluation has been arranged so that no one below the rank of Captain gets the “Outstanding” rating, by the metrics used. And anyone below the rank of Captain has to have it signed off personally by the Warden to get an “Outstanding”. Which does not happen.
At the rank of Captain and above, the metrics are far more forgiving and oriented to shuffling papers not dealing with felons. Thus, any money appropriated goes to Captains and above. If you look at the ratings, Colorado has the finest prison administration in the country, with everyone rated “Outstanding”.
The metrics for Wardens, one year, were based on how few Class I violations of the Code of Penal Discipline the prison had. Class I violations are things like murder, rape, assaults, possession of drugs, etc. The response of our various administrators was to have the Duty Officers, who were on call for the Wardens at night and weekends go through the reports before they were formally logged, and destroy any Class I disciplinary reports except for Escape and Murder charges which could not be covered up. Which incidentally meant that the inmates got away with anything for that year. But the Wardens got their bonuses. And the line staff had to pick up the pieces.
After a decade and a half of this, Colorado prison line staff are paid a fraction of what their Federal counterparts are, and well below what jail deputies are paid in a lot of counties. And they wonder why they have trouble recruiting and retaining good staff. But apparently, as I said, they have the best administrators in the country.
Subotai Bahadur
Curious if there is also an iatrogenic effect from reporting or blogging.
For instance: I’ve used the Target phone several times, and have always received prompt help. I know several family members who have expressed satisfaction with it.
As a part of the overall systems design for a large retailer, designers probably realize that there won’t be a concerned and caring manager standing by to take charge of each individual phone call and assign a worker responsibility for each call as it is received. The company is too big for that, and if they could afford that many extra managers, it would be more efficient to simply hire more line workers and place one by each phone. But they can’t.
So, they set up a timed-response system to get the help out to the customer quickly. I, the customer, really don’t care what has motivated the employee to get to me so quickly. Whether I am shown how to find light bulbs quickly by an individually-motivated, enthused and diverse workforce, or by an employee who gets to me quickly only because his response times are recorded and used in his evaluations, is immaterial to me. I just want light bulbs.
But you’ve written this post that makes the quick response sound sinister and unfair and wrongheaded, and you’ve done it using a voice that seems to say “hey, Target, here’s some help for you”. So, some previously happy customers might now look at the quick help with a cynical eye.
Would this not be a case, then, of an iatrogenic blog post?
BobbyB…not sure what the size of the company has to do with it. Local store employees are managed by local store managers, and the ratio need not change whether there are 3 stores or 300.
Your experience seems to have been better than mine; it is possible that there are regional differences. I don’t go to Targets all that often, but who I do, sometimes some of the CS phones haven’t been working at all and generally when using the “operator” option, the person at the other end hasn’t known where stuff was. (I was a TGT shareholder until I sold a year and a half ago or so, and hence have paid more attention to operations than I normally would in a store.) The system has come across as pretty unmanaged.
As Phwest said, CS metrics are like sales comp metrics; they have to be watched and tuned.