I’m only kind of a football fan. Mostly that’s because my football team left town (San Diego Chargers) after regularly snatching defeat from the jaws of victory in the fourth quarter.
But the recent NFL scouting combine was fun to watch and for me thinking about predicting success.
I’ve written and spoken about the research Ashish Nanda did at Harvard related to high performers. The short version is that while studying high performers in the financial sector (investment banking), they saw that when these high performing “free agents” changed organizations, their performance dropped. And not just for a year or two – but for five. You can read more about some of the research here.
In essence, predicting high performing success is hard work and is more than just looking at some individual performance metrics.
But isn’t this exactly what the NFL combine does? And how do those results turn out?
Here are some highlights:
- Of the 1300 draft picks from 2000-2012, more than 35% were out of the NFL in 2 years.
- Roughly 35% of draft picks never start a game.
Wouldn’t it make sense, given how much money is spent in each draft, to figure out a way to reduce the risk of these decisions?
People have always used data to make decisions. Now artificial intelligence can help (in the form of machine learning).
Here’s the good news – a model was created (and trained using machine learning and a lot of data) that mitigates the risk of making poor choices, while helping suggest good candidates for QB success.
And the details behind the model are also available for anyone to read – as it was a proposal for a conference presentation.
But here’s the challenge. These models are not perfect. While they help you avoid making many mistakes, they also can hide the opportunities in finding the next Tom Brady.
If your approach, in predicting high performers, is driven by risk mitigation, you open up yourself to the worst kind of poor performers – those whose history isn’t a predictor of their future performance.
As I watched the combine, I heard a quote from a coach who, in referencing a prospect, said they were the person they most liked on the field and didn’t like the most off the field.
So how do you choose? Based on actual performance on the field? Based on your gut?
When I hire people, of course I look at their history. I look at their performance. But I also put them into situations with their future team to see how they perform. And I even try to hire people who have history with each other, to mitigate the challenge of building a high performing team from strangers.
Success at work is unlike the NFL. We aren’t running short highly-choreographed plays with refs and rules.
So how do I predict individual success as I make new hires?
I look at how a person has handled change, in their professional history. It could be little or big change. But the chaos that comes with change often reveals a lot to me about how a future employee will experience and handle the stress of change.
I look at how a person has handled growth, in their professional history. Are they repeating the same year or two at work, over and over? Or are they constantly learning?
I look at how a person has handled relationships, in their professional history. Have they developed close friendships with the people (at least some) they’ve worked with? This helps me understand and predict how they might navigate joining a new team as well as help me predict if they’ll stay for a while.
I find the NFL combine a lot of fun to watch. And I’m a big fan of data, machine learning, and making critical business decisions.
I don’t think there’s a right or wrong strategy for using data when it comes to predicting success.
I think failure only comes when we have no model at all – when we simply get excited and use hope as a strategy for predicting individual success.