You’re General Shinseki.
What happened to him could happen to you. Here’s how to prevent it:
But first, let’s get two facts straight.
Fact #1: The scandal at the VA isn’t that the VA provides awful care to veterans. Veteran satisfaction with VA care is on a par with private-sector care. An acquaintance who heads a chapter of the Vietnam Veterans Association and has been involved in the current inquiry confirmed for me that care quality isn’t an issue.
Fact #2: Nobody in the VA delayed care or treatment. Veterans seeking care were scheduled into the earliest timeslots available.
The “scandal” is that managers throughout the VA required staff to fudge the numbers to make it appear the agency was meeting its required service levels.
The VA’s leaders, from Shinseki on down, didn’t act on chapter 3 of the KJR Manifesto. They didn’t, that is, avoid Metrics Fallacy #4, which is, in case you need reminding, extending organizational metrics to individual employees. It’s a fallacy because the moment you do you won’t be able to trust your numbers any more.
What was true for Jeff Skilling and Ken Lay at Enron is just as true for a low-level manager at the VA: When your performance is gauged by numbers, you have an incentive to fudge the numbers, which in turn makes the numbers useless for gauging organizational performance.
If you lead a large IT organization I’d bet good money it’s happening to you right now.
Start with time tracking. Employees all know that if the numbers show they’re under-utilized they’re more vulnerable to the next round of layoffs. Think they don’t allocate some of their open time to various categories of doing something productive?
Of course they do.
Now think about Agile. Many Agile variants use some form of backlog management, the “backlog” being the project’s to-do list of desired new or improved system capabilities. These are usually described in terms of “user stories,” which describe what’s needed.
Each user story receives a consensus degree-of-difficulty rating from the Agile team. Agile teams become very good at this, which in turn means Agile projects forecast delivery more accurately than traditional waterfall projects.
It has to be tempting to use weighted user-story development time to rate developer performance.
Resist the temptation. Succumb and here’s what you’ll get: padded degree-of-difficulty estimates, slower development because developers will live down to their inflated estimates, and inflated performance numbers, because of the same padding.
Then there’s the Help Desk. Help Desk managers have a lot in common with VA scheduling managers, in that the amount of staff time available to resolve reported incidents is often quite a bit less than the time needed. Think your Help Desk staff don’t quietly close incidents they’ve touched but that aren’t really resolved, to make their close rate look better and their incident backlog smaller?
Which brings us to the most astonishing aspect of this whole sorry mess: General Shinseki apparently acted like a manager, not like a leader, and not a very good manager, either.
Shinseki, that is, relied on reports, his 12-level chain of command, and extensive time spent with the VA’s regional medical directors.
None mentioned the metrics-fudging. Why would anyone expect them to?
Generals are supposed to know that if they want to know what’s really going on, they need to talk to the soldiers. I’ve encountered no hints that Shinseki did much of this.
Look, fudging management reports to make performance look better is a time-honored tradition in the world of organizational dynamics. This isn’t a scandal in any meaningful sense of the word.
If there’s a scandal, it’s that the metric that mattered most … a metric on which lives depend if it’s the VA, and your relationship with the business depends if it’s the Help Desk … was jeopardized by the insistence on measuring individual performance. That metric? Demand/Capacity. So long as it’s enough less than one to cover day-to-day variations in demand, veterans will get the care they need when they need it and users with failing tech will get the fixes they need.
Turns out, in the VA it’s nowhere close.
Which in turn reveals Shinseki’s true failure.
The VA budget has increased significantly during the past two administrations, yet somehow that budget increase hasn’t turned into sufficient capacity.
With money to spend, Shinseki and his predecessors failed to make sure it was spent hiring and retaining doctors and nurses. That’s a scandal.
I trust the parallel is clear.
Yours is the first analysis I’ve seen that actually makes sense of this tragic mess.
Tragic because so many died unnecessarily because no one was tasked to do the kind of management analysis you did. Or, more likely, was actually capable of doing it. And, tragic, because by all accounts, Shinseki is a courageous individual of extremely high personal integrity, but clearly didn’t have “the particular set of skills” to really know how to start to deal with the VA’s problems.
Good one, Bob. Okay, right, they’re all good; but, this one stands out as being extra goody. Keep the good stuff coming.
…..Michael
Our parent organization keeps trying to force Remedy–a helpdesk software ‘solution’–onto us. They claim it will issue tickets and provide metrics on numbers of calls, and time spent.
Our stance is that we don’t care. Our users want the problems fixed and don’t care about counting them up in monthly reports. And the management analysis is that if we improve the system based on user calls, then over time the calls should get harder and harder and take longer and longer to solve (which is what our informal metrics tend to show).
Many times, even if you got accurate numbers, they don’t actually tell mgmt anything useful.
I was waiting for this post. I knew it was coming.
The maddening part is that the people who run the VA, from top to bottom, all the way to the President himself, would have know this to be the issue if they had any real knowledge of management and metrics.
Bob, you have covered how metrics get skewed and how to find the right metrics to measure for over a decade now.
While there is no chance at all that you’ll get the nod from Obama to run the VA, it should be people like you that are tapped to do this kind of work. Picking managers because of there political connectedness strikes me as being as useful as picking people for IT positions because of the sports teams they cheer for. How about they pick someone with the competence to solve the problem that exists for once?
Regrettably, running for office on competence appears to be a losing strategy. Everyone in the loop, from advisors to the commentariat, think in terms of policy (intentions) and the ability to get legislation passed. Running the joint?
To be fair to president Obama, he seems to have paid more attention than most to appointing competent people. And Shinseki was a highly regarded general – appointing him to head up the VA seemed reasonable.
Turned out he was the wrong person for this job. As a point of comparison, most CEOs brought in for turnarounds usually fail too. The difference: they can usually find a buyer after they do.
After reading your articles for a number of years, I had your analysis figured out before I was half way through. You’re a good teacher. Somewhere there was an article, after the Clippers announcement, in which a U. of Colombia professor said Ballmer will get called on many double dribbles if he takes over.
My operations clients are generally terrible at measuring anything, but of the ones who do, on-time performance is a pretty common priority. Unfortunately, they almost all lie to themselves (and their customers) by claiming bizarrely high numbers. One wholesaler of specialty products claimed 98% on-time delivery but complained about late deliveries from their biggest suppliers. And they didn’t stock many of the specialty products, so inventory couldn’t buffer the shortages. When asked how they measured on-time delivery, they explained that if an order wouldn’t be in stock to ship when they originally acknowledged it would, they’d correct the shipping date. And they were supposed to call the customer to let them know. And they often did.
So, do their customers think they have a 98% on-time rating? Their salesforce says it’s the thing their customers complain about most.
On-time: “You keep using that word. I do not think it means what you think it means”
While no individual VA employee wrongly denied care to a vet, the end result was the vast denial of care. There was little opportunity for Shinseki (or Congress) to fix the problem, because the existence of the problem was effectively covered up through systematic deception by middle management.
Those who lie on official documents should be fired. Those who direct them to lie should be fired as well.