A number of articles on the Internet explains the difference between lagging and leading measures. I gave some explanations as well; my latest was using a golf metaphor:
- Knowing the score or the distance between the ball and the cap is easy (lagging measure)
- Knowing and controlling the parameters of golf swing that lead to an excellent score are hard (leading measures)
By just telling this we do as little as defining things. Now everyone understands that we need both lagging and leading measures, and that leading measures are hard to find. Knowing the problem does help us a little, but still, there is a need for some specific recommendations about finding leading measures.
Let’s talk about a toolkit that can help us to come up with excellent leading measures.
In order to be consistent, I’d like to start with the nature of measurement itself. Before we were talking  about such steps as observation and quantification that one needs to follow in order to come up with a metric, but in the business world most of the measures are actually simple calculations. The measurement is probably as old as science itself, still in the business we often forget about scientific nature of measurement and replace it with simple calculations.
Scientific research implies a certain level of uncertainty about a subject of the research and the goal of measurement is to reduce the uncertainty. If we have a look at business measures from this point of view then we will be really disappointed. The majority of “measures” are actually calculations  or are based on the information obtained directly from the subject of the research.
Measurement vs. calculation
Let’s take an employee turnover as an example. An HR manager calculates turnover by dividing the number of employees who left the company over a period by the average number of employees that were in the company during this period.
Is there any magic about this? All of the variables are available in the company’s ERP system and are simply divided. We could call it Big Data, but the idea is as simple as running two database queries to get two variables and divide one by another.
There is nothing bad about calculating something that might be insightful for HR, but let’s not call it measurement, as it is just a calculation.
Social science measurements
Another “measurement” paradigm uses the method of social science. I’m speaking about surveys. The idea is to ask the persons that are in the focus group (the subject of the research) to share their opinion about something. One supposes that well formulated questions will minimize the risk of receiving wrong information and making wrong conclusions as a result.
It is not a calculation, but still we are doing only what social scientists can do – ask people directly about their opinion.
Traditional scientific measurement
Traditional scientific measurement method is about observation. Scientists cannot ask nature directly about laws of physics or its “opinion” about gravity. Instead, they observe, they come up with hypothesis, and they test them to see if one worked or not. It is always about the reduction of uncertainty and almost never about precise calculations.
Different ways to find measures
A practical conclusion from the ideas described above is that we always have several measurement options.
- We can “calculate.” It is not a measurement, but still it gives us a number
- We can communicate with the subject of the research (with our customers, prospects or employees), for example via surveys
- We can observe and do conclusions (we are not limited to observing our own business only or the present day only)
- We can formulate a hypothesis and test it under artificially created conditions
Let’s have a closer look at these types of measurement. I’m now curious about how useful they are for finding leading measures.
We could calculate satisfaction rate by calculating the number of repeat purchases. This is not a measurement, it is a calculation, but it still reduces the level of uncertainty about the subject in the sense that it allows managers to see the problem from a different perspective.
- With calculations we could normally generate excellent lagging measures, but not leading ones
Here we are talking about “footprints” discovery. We suppose that satisfied customers are those who make repeat purchases within 2 months. That’s great to know, but we have no idea about what leads them to a repeat purchase.
We could calculate some customer activity by taking some variables, but sometimes it is easier to conduct a survey by asking customers about their preferences and decisions. When I say that it is easier, I mean that the procedure of measurement is easier itself, but the formulation of survey questions is always a challenging task.
- What type of measures will we have as a result? It depends a lot on the kinds of questions we ask. With some skills it is possible to come up with a hypotheses about driving factors of customer decisions and to find out leading measures.
Observe and conclude
In this case we don’t focus on the numbers and we don’t ask customers directly, we just observe what they do and make appropriate conclusions. You might install a special script on the website that will show you the most clickable zones. You can observe the process of customer choice in the physical shop.
As I mentioned above, we are not limited in the time or in the places where to use this method. We could make some hypotheses basing on historical values of our company, as well as comparing our ideas with other companies that have faced similar challenges.
Artificially created conditions
A measurement by observation as described above might be as difficult as finding black holes in the space by tracking light behavior. We are not pretending to do a space science in the business, and we can artificially create some conditions and see what will happen.
Take a/b tests as an example. Instead of asking customers for their opinion, or silently observing their behavior, one could create two options and see which one works better.
- These two measurement methods actually give us more opportunities to find excellent leading measures. Under artificially created conditions you can test business hypotheses and get closer to the drivers of performance excellence.
All these approaches might give contradictory results
It is not a real business case, but I think you do agree that we are faced with situations like this very often:
- Our CRM system gives us a number, but we don’t know how we can change it.
- We do some survey and sometimes we get some ideas, but they might not be coherent with real customers’ actions.
- We obverse and do some conclusion about what might work
- We try various options and it appeared that something unmentioned works better
For this example, I would not try to come up with a leading measure yet. It is obviously that the company needs to do some more tests or conduct more specific surveys to understand the challenge better.
- Finding leading metrics is actually about finding a solution that works
Finding excellent leading measures
I think now it is obvious how one can find a better leading measure. We need to have a balance between various measurement methods. We need to combine the results of calculations, the survey results, the results of observations and the results of tests under artificial conditions.
Some safety rules need to be followed:
- Have your business context in mind. We don’t want to measure something just because we can measure it.
- Don’t believe people who tell you that it is impossible to measure, as it is possible to quantify and measure anything .
- If you cannot quantify something then you need to decompile the subject into measurable pieces
- Combine various measurement and research methods like calculations, surveys, observations and a/b tests.
Are you trying to find a good leading measure? Post the details in the comments and probably someone will suggest some good survey questions to ask, observation method to use, or an a/b test to do.
- ^ Aleksey Savkin, The difference between quantification, measure, metric, and KPI, January 2014, BSCDesigner.com
- ^ Douglas W. Hubbard, “How to Measure Anything: Finding the Value of Intangibles in Business”, 2007, Wiley; 1 edition