The Window Tax
In 1696, the English government came up with a clever way to raise money. Instead of counting people, land or livestock, it counted windows. The more windows your house had, the more tax you paid. Simple. Unfortunately, people quickly discovered an even simpler solution. Brick up the windows. Why pay tax on a window you don't actually have? Soon, houses were being built or altered with fewer windows. Existing windows were blocked up and houses became darker. The government had found a way to measure wealth. People had found a way to avoid the measure.
It is a very old management problem. Today we call it a KPI. Most organisations use targets because they want to improve something. Productivity. Response times. Customer satisfaction. Complaints. Staff turnover. Costs. The intention is usually sensible. Pick something you can measure and use it as an indication of whether things are getting better. Then the target becomes important. Then very important. Then the target becomes the thing. And that is when things start to go wrong.
Suppose a customer service team is given a target to answer calls within two minutes. The target is designed to improve customer service, so staff learn to keep calls short. The target is hit. Customer service may not have improved. Customers may simply be getting quicker conversations. Or a council sets a target to reduce the number of complaints. Staff discover that the easiest way to reduce complaints is to make it harder to complain. The numbers improve. The underlying problem doesn't. A target to reduce sickness absence can encourage managers to pressure people back to work. A target for recruitment speed can encourage shortcuts. A target for closing cases can encourage people to close cases before the real problem has been resolved.
The organisation can become very efficient at doing the wrong thing. Time is spent optimising the target rather than improving the service. Staff learn which numbers get attention and which problems can safely be ignored. Managers receive reassuring reports while customers experience something quite different. And when managers complain that staff are gaming the system, they can miss the obvious question: who designed the game?
This is the uncomfortable lesson of Goodhart's Law: when a measure becomes a target, it can stop being a good measure. A KPI is only a proxy for the outcome you actually care about. The danger comes when managers forget that. People are not stupid. If their pay, reputation, appraisal or job security depends on hitting a number, they will pay attention to the number. That is not necessarily gaming. It is rational behaviour. The management failure is assuming that people will optimise the outcome when the organisation is rewarding them for optimising the measure.
The answer is not to abandon KPIs. Without measures, organisations can end up managing by opinion and anecdote. The answer is to manage the measure as carefully as you manage the performance. Before introducing a target, ask four questions: What behaviour will this target encourage? Could someone hit it while making the underlying problem worse? What other evidence would tell us whether we have achieved the real objective? And, most importantly, if people do exactly what this target rewards, will the organisation be better off? If the answer is no, change the target before people change their behaviour.
The Victorians eventually got rid of the window tax. We have been creating new versions ever since. The difference is that modern managers have dashboards, spreadsheets and performance reports to tell them exactly how well their targets are being met. Which is very useful if you want to know how many windows have been bricked up. It is less useful if you want to know whether the house is still fit to live in. Because the most dangerous KPI is not the one nobody hits. It is the one everyone hits while the organisation quietly gets worse.
