When KPI’s drive the wrong behaviour
There is an amusing story I like to share during training.
A few years back, the city of Mumbai had an issue with a rat infestation. The rat population was growing rapidly and, of course, causing all manner of health concerns. Then a government official had a great idea. Mumbai also has poor people, so simply pay people to kill rats – deliver X numbers of dead rats and receive X number of the rupee. Brilliant.
It worked beautifully. The rat population declined rapidly. But after a few months, it started to rise again and the number of dead rats delivered for payment rose as well. No-one could understand why. Investigations found the answer – people were breeding rats. Makes sense, as rats breed quickly and they were proving a useful source of income to struggling folks.
The next true story is not amusing. A pizza company had an objective “Deliver Superior Customer Experience,” and a supporting measure of “95% of pizzas delivered within 15 minutes.” This was based on research that found that customers wanted their food hot and delivered quickly. Makes sense, and to encourage the delivery of this, customer experience outlet managers’ bonuses were based largely on this KPI.
One day an outlet had an issue with its ovens and there was a panic that the target would not be achieved. The manager told a delivery man to get on his motorcycle and hurry up and get to the customers’ homes to hit the target. The young driver drove quickly, crashed and died.
Both these amusing and tragic stories deliver the same message. Be careful what you ask for when creating KPIs and setting targets, and most notably when incentive-compensation is tied to performance. They might just encourage “rational” behaviours that could be either positive or negative.
Dysfunctional behaviour is common
From my experience, dysfunctional behaviours (which are simply “rational” responses) triggered by a KPI are far from uncommon. It is well known (and I have seen this) for a manufacturing plant to be set a target to reduce reported injuries, and for the target to be reached simply by only reporting serious injuries (that can’t be hidden). Performance does not improve, but the target is hit. There are many similar examples from all sectors, industries and functions.
Brainstorm the positives and negatives
When I work with organizations to select KPIs and targets, a step I always include is to get people to think about the behaviours the KPI might drive. I simply ask them to brainstorm and write down all the positive behaviours that might be encouraged and then the negatives. When done, we then discuss how to best encourage the former and mitigate the latter. Sometimes the risk of dysfunctional behaviour is so great that the KPI has to be rethought or abandoned – a simple exercise that can deliver a lot of benefits and save a lot of heartaches.
…write down all the positive behaviors that might be encouraged and then the negatives.
Changing the intent
Sometimes negative behaviours can happen as the KPI transitions from being designed to report.
I worked with one government organisation that had a KPI for the Finance Department of “90% of invoices paid within two months.” This was something of a stretch for a government entity in that country. So, when I reviewed the finance scorecard I was surprised to find the colour was green. It was being hit. This made little sense to me, as in my conversations with suppliers, a common gripe was that it took up to eight months to get paid. Clearly something was remiss.
Auditing the KPI found that although the original intent was payment within two months of receipt, finance had changed this to two months from final sign-off, which in this very bureaucratic organisation took six months. Again, performance did not change, but the target was hit. No need for exception reporting here.
…performance did not change but the target was hit. No need for exception reporting here.
So, when designing KPIs and targets think about the rational behaviours (positive and negative) that might be encouraged and plan accordingly. But also ensure that the original performance-enhancing intent of the measure is not changed (oftentimes surreptitiously) during implementation. Indeed, a regular audit of the Balanced Scorecard is good practice (and when this is resisted by managers it’s a strong indication that something just isn’t right).
Finally, be particularly careful when bonuses are linked to KPI target achievement. There’s an old adage that says “What gets measured gets done. What gets rewarded gets repeated.” Be careful you don’t simply end up rewarding more rats.
I find your article interesting and true. Our government agencies should look at their KPI’s and see what poor behavior they (or we the people) are rewarding.