In statistics, a proxy or proxy variable is a variable that is not in itself directly relevant, but that serves in place of an unobservable or immeasurable variable. In other words, sometimes there’s not a good way to measure directly what you need to measure, so you use another metric you can measure that correlates with the one you can’t.
Let me provide a very specific example: Leadership is difficult to measure. It’s not like a gallon of water. You can’t pour leadership in a bucket and look at it and say, “Yup, we have ourselves a nice gallon of leadership.” Because we can’t measure it directly, we assess it in terms of outcomes. This is why we use proxy metrics like stock price, return on equity, net promotor scores to gauge leadership performance. None of these are perfect, but one way or another they get the job done.
You can’t pour leadership in a bucket and look at it and say, “Yup, we have ourselves a nice gallon of leadership.”
There has been much written about how effective leaders manage worksites that have fewer accidents and injuries. This makes logical sense: good leaders care about their people. When they care about those closest to the work, they put in systems, create a culture, and provide the resources to ensure that resilient defenses are in place to mitigate and sometimes even avoid terrible unplanned events.
This leads us back to the problem of measuring leadership effectiveness.
If good leadership equals less people getting hurt, let’s whip out those proxy leadership metrics and go to town! If it were that easy, 100 people wouldn’t be dying on the job every week in the United States (sadly, this metric is direct and it has stubbornly not materially gone down in the past ten years). Unfortunately, traditional proxy metrics like stock price or return on capital or net promotor score have not proven to be good predictors of unplanned incidents in the workplace. There are good reasons why they are not effective, but that is another post.
A great deal of research has been undertaken to develop a good proxy metric that links effective leadership and unplanned incidents in the workplace. Paul O’Neil famously turned around Alcoa both from a shareholder value and safety perspective doing this very thing. Dupont, DEKRA, Virginia Tech, and many, many others have done significant work in this area. In most cases, their proposed solutions end up measuring safety climate or culture in one form or another. Usually this involves a survey (or two or three) and an expert who tells you what the survey says. Guess what kinds of questions are on the survey. Yes, you guessed it. They are designed to indicate or provide a proxy for leadership health. If you have the money for these services and you can do them frequently, there is nothing particularly wrong with this approach. You just have to accept the obvious caveats: they’re not cheap, you can only survey folks so often, and you ultimately have to do something with the information and that action is sometimes harder than you might imagine when it comes to leadership.
But there’s another option that is a lot more cost effective that doesn’t call for implementing a new process or system or asking your teams to do anything they are not already doing: Monitor and measure the quality of conversations between front line leaders and those closest to the work when they have their daily planning meeting. Yep, that 10-minute meeting. Just to be clear, I said measure the quality of conversation. I did not say measure if the conversation happened or that someone filled out a form.
Here’s why these conversations are such great proxy metrics for good leadership.
Good leaders and executives retain, train, and encourage front line leaders to have effective daily planning conversations with those closest to the work. Period. If you have a leader who leads a high-quality daily planning conversation, he is very likely to have other effective conversations throughout the day, have a plan for the work that needs to get done, have the trust of his people and care enough to make sure they have the resources they need to protect themselves from the inevitable unplanned event that may or may not end up in a serious injury.
Don’t believe me? Fair enough. We recently completed a study looking at 74 projects and five thousand or daily planning conversations over a two-year period. Some projects had incidents where someone got hurt and some were injury free. The projects where the daily conversations scored below a threshold of 2.5 on a scale of 0-3 for Quality of Conversation were nearly 4X as likely to have an incident than those with scores above 2.5. (Click here to read more about how you can measure conversations and how we get these scores.)
Although this is a relatively small sample set, the findings are very compelling. We expect to learn more as we continue to sift through the tsunami of conversation data our customers are now sharing with us. Our research is also extending beyond safety and looping back around to the impact of conversations on schedule, profit, employee retention and rework.
Our position is that leadership health is single most important metric you use to order who needs your attention and resources most.
Predicting an unplanned event in the workplace like a serious injury or fatality is very difficult. Some would even argue it isn’t possible. We don’t have a dog in that hunt. Our position is that there are some worksites that are more at risk of something really bad happening than others. On high risk worksites where humans are still required the risk of human error and associated catastrophic consequence never goes away. Our position is that leadership health is single most important metric you use to order who needs your attention and resources most.
To this end, if you want a good proxy metric for leadership and you want some evidence that your teams are operating with incident preventions systems that are robust and resilient, we strongly suggest you listen to their daily planning conversations and consider what is happening to enable the ones you want to hear and is holding back the ones you want to change.