Even since I studies Statistics, and Six Sigma principles I got interested in Metrics. I had studied a subject on Instrumentation and Control in my under-graduation (Mechanical Engineering). But it was more about what different tools you use to measure, to take readings in conducting experiments. It was not about Metrics – the purpose and principles of setting right Metrics.
The purpose of Accounting (Balance Sheet, Profit and Loss Statement, Cash Flow Statement) is to “measure” how the business is doing. You need a lot more analysis and ratios to understand the health of business; however it starts with the basic activity of recording, and reporting of transactions – i.e. by measuring. The same is true for everything.
Harvard Professor Clayton M. Christensen wrote an article in 2010 titled “How Will You Measure Your Life?“. It was later published as a book with same title, which received a lot of attention and applaud.
Without getting into details of the book, the point I want to emphasize is on “measuring”.
That which is measured, will improve.
Actually this is not entirely true. Here is the precise argument.
And this view causes all complications and trouble. Everything in life comes at a cost – it could be monetary cost (tangible, countable), or non-monetary costs such as Opportunity Cost, Social Cost, Psychic Cost etc.
Thus the dilemma is not only about “measuring to improve”, but “at what cost?”. Cost benefit analysis of measure becomes an important part in designing the measure. Let me give an example. In 100 meter sprint, the winner of 1920 Summer Olympics clocked 10.6 sec, the runner-up 10.8 sec and the second runner-up 11.0 sec. The time recorded was only up to first decimal place i.e. 1/10th of the second was “the least measure”. Now we can record up to two decimal places i.e. 1/100th of a sec. And this comes at a cost. We need technology, support staff etc. to measure that level of accuracy.
Can we assume that we cannot record time up to 1/1000th of time? Maybe not. We probably “can” record a time such as 9.975 sec; but “at what cost?”. And more importantly: is it required? If the winner is completing the race in 9.58 sec (Usain Bolt!) and the second-best is taking 9.62 sec, why do we need accuracy up to 1/1000th of sec?
The same is true about any other measurement. John Maynard Keynes, the famous Economist and Nobel-laureate once said: “It is better to be roughly right than precisely wrong“. You should not “measure” something with lot of accuracy just because you have the tools and techniques available. The important question is: Does it serve the purpose for which you want to measure?
However, lot of times people confuse measure with target. Goodhart’s Law summarizes it well.
Increasing employee headcount to 100,000. Having 100 offices across the world. Being on cover of TIME or Forbes magazine. Are these targets or measures? You may reach 100,000 employees or 100 offices by acquiring another company. Then what? What you do with employees and offices should be a goal and not the 100,000 or 100.
And yet organizations often set wrong KPIs. A CEO of $500 million IT Services Company would be given a target of achieving $1 billion in 2 years. The $1 billion then ceases to be a measure and becomes a target. He would tell his subordinates to “hire more people” to achieve the $1 billion target. “Hiring more people” then becomes a target for his subordinates. The hiring target would then be broken down into X% of laterals and Y% of freshers. The campus hiring team would go to campus and select Y% from ANY stream, because Y% was set as the Target! The selected freshers from unrelated streams (Chemical, Mechanical, Petroleum?) need to be trained. So you build training facilities, university like campus, run training institute like L&D programs – because training Y% has become your target!
It’s a chain reaction…
I will write in next blog about few other aspects related to implications of Metrics and its power to change behavior.