Pete Carroll must forever live with the consequences of a good decision gone bad.
On Feb. 1, 2015, Carroll’s Seattle Seahawks trailed the New England Patriots 28-24 with 26 seconds left in Super Bowl XLIX. Seattle had the ball second down and goal on New England’s one-yard line with Pro Bowl running back Marshawn Lynch in the backfield. New England had the worst record in the league that year of allowing opponents to score within two yards of the goal line. All the stars were lined up to give the ball to Lynch and let him barrel into the end zone.
But to nearly everyone’s astonishment, Carroll called for quarterback Russell Wilson to pass. The throw was picked off by Patriots’ cornerback Malcolm Butler, who fell on the ball, enabling New England to run out the clock for a stunning and unlikely win.
Sports pundits were merciless the next day. “Worst play-call in Super Bowl history,” trumpeted the Washington Post. “Pete Carroll botches the Super Bowl,” wrote ESPN.
Statistically, though, Carroll’s call was sound and even brilliant, notes decision science expert and former world-class poker player Annie Duke. NFL teams had thrown 66 touchdown passes from the one-yard line that year with zero interceptions. Throughout the history of NFL record-keeping, the chances were 98% that the play would have resulted in either a touchdown or an incompletion, either of which would have benefited Seattle.
The uproar surrounding the head coach’s decision is an example of what Duke calls “outcome thinking,” or the assumption that a decision that leads to a negative outcome is, by definition, a bad decision. In her recent book, Thinking in Bets, Duke notes that outcome thinking compounds poor choices on two levels: It dissuades us from making sound future decisions while reinforcing bad decisions that turned out well thanks to a lucky break.
Uneasy hindsight
We all know examples of outcome thinking: hiring the hotshot CEO who turns out to be a tyrant in the workplace or choosing a promising-looking vacation property on Airbnb that’s infested with mice. When such bets don’t pan out, we tend to blame the board of directors who hired the executive or the booking agency that we’ll never do business with again. That’s although both outcomes were anomalies that aren’t likely to happen again.
Outcome thinking undermines the data-driven decision-making culture that is necessary for digital transformation. We have more information at our fingertips than ever, but seat-of-the-pants decision-making that has been solidified by years of habit persists. A Business Application Research Center survey reported last year that nearly 60% of business professionals said managers at their companies base at least half of their decisions on gut feel or experience.
Not all decisions demand rigorous analysis, of course. There’s a lot less at stake in choosing what to order for lunch than making a marriage proposal or deciding whether to bet $3 million on a startup. The greater the risks of a bad decision, Duke asserts, the more important it is to rely on data.
Getting access to that data is easier than ever. Cloud computing has democratized data warehousing, making it possible for anyone to tap into the power of analyzing massive data stores, which are themselves available as cloud services. Machine learning algorithms, which are essentially probability engines that make recommendations based on correlation, are proliferating and becoming easier to use.
Playing the odds
Humans, however, are still coming up the evolutionary curve. Many executives like to draw analogies between business and chess, but a more accurate comparison is to poker, Duke says. A chess player is in full control of his or her destiny and can only lose by making mistakes. In contrast, poker players live in a world of uncertainty where even a champion can lose to a novice in any given tournament thanks to a few lucky breaks. Winning over the long term requires understanding the odds and making good bets repeatedly with the understanding that they won’t always pan out.
The strategy Duke recommends is to be specific about the data that undergirds critical decisions and our confidence level in them. Instead of using terms like “significant” or “huge,” cite the known facts, the calculated probabilities, and your confidence level that the choice is the right one, even if that confidence is an educated guess.
When that happens, “Making better decisions stops being about wrong or right but about calibrating along all the shades of gray,” she writes. Relying on data and statistical probability gives everyone a clear foundation for making decisions and shared responsibility for the risk of failure.
Winning in business is rarely an all-or-nothing proposition. Walmart has less than a 10% share of retail sales. Decisions that are successful 70% of the time yield enormous profits. Just remember, Duke writes, “an event predicted to happen 30% to 40% of the time will happen a lot.”
Next read this:
Copyright © 2021 IDG Communications, Inc.
