In the spring of 2022, after a grueling 58-loss season, the Oklahoma City Thunder sat near the bottom of the NBA standings. Then something remarkable happened: The franchise started getting better, fast.

The following season the team won 40 games, blowing past Vegas projections by more than 16 wins. The next year it jumped to 57 wins, exceeding expectations by 12.5. A year later the team reached 68 wins, another 10.5 above forecasts. In just three seasons the Thunder surged from the bottom to the top of the league. Despite having the youngest roster ever to earn the number one seed, the team finished the season as the NBA champion.

How does a team improve so quickly? That question isn’t just about basketball. It’s about how any group can learn, adapt, and succeed in an age of accelerated change.

During the past few years, my colleagues and I have been investigating a question at the heart of organizational success: What do the best teams do differently? To find out we surveyed more than 6,000 knowledge workers across a wide range of industries—from finance and law to healthcare and technology—and collected detailed data on how their teams set priorities, make decisions, and collaborate.

To identify members of high-performing teams, we had respondents first rate their team’s effectiveness and then compare its performance with that of other teams in their industry. (Prior research has shown that people’s belief about their team’s ability is one of the best predictors of real-world results, making it a reliable proxy for team performance.) Workers who scored their team 10 out of 10 on both measures were designated members of “superteams,” allowing us to compare their behaviors with everyone else’s.

Among those superteams, the formula for high performance looked remarkably similar. Superteams share three key strengths: (1) They get more done by managing time, energy, and attention more efficiently; (2) their members actively make one another better; and (3) they’re constantly building new skills and improving over time.

It is that last strength—improving over time—that the Thunder has clearly mastered. Its rapid ascent over the past few years is not an anomaly. It’s part of a pattern that has defined the organization since it moved to Oklahoma City from Seattle, in 2008. After its arrival, the team entered a full rebuild, prioritizing long-term progress over instant results. Yet in just four seasons it reached the NBA Finals, in 2012.

But success in the NBA rarely lasts. Star players age, salaries rise, and league spending rules push even the best teams to rebuild. What makes the Thunder organization exceptional is that it pulled off the same improbable climb a second time: not only in its early years but again in today’s faster, three-point-heavy game, both times with one of the youngest teams ever to contend for a title.

How did the Thunder do it? By applying strategies that are surprisingly common among superteams—strategies that any group can use to accelerate improvement in an AI-driven world where a team’s ability to learn and adapt has never been more crucial.

How to Drive Continuous Improvement

At the center of every learning culture is a leader who makes growth a daily priority. Our research uncovered seven practices that top leaders use to foster continuous improvement. The same principles that built a championship team can help any leader create a team that keeps getting better.

1. Run more experiments.

Superteams evolve by trying new approaches, even when things are going well. By contrast, average teams often stall as soon as they find something that works. They stop asking questions. They stop trying new things. And eventually they fall behind.

In our research, superteams reported experimenting nearly 50% more often than their average peers did. Those experiments can range from small trials, such as A/B website testing, to much bigger ones, such as selling a new offering before it exists. The key is to always be working on something that yields new insights, learning, and growth.

You can see that mindset reflected in today’s most innovative companies. Meta, Netflix, and Amazon, for example, run thousands of experiments a year, testing everything from product features to new business ideas.

But experimentation doesn’t always spread on its own. At Spotify leaders noticed the same pattern we saw in our research. A few teams were experimenting; most were not. So leadership set out to increase experimentation at scale by creating an internal platform called Confidence, which made testing easy and transparent. Any team could launch an experiment, share results, and learn from others’ data. Within a year more than 300 teams were running tests that shaped everything from playlist design to podcast discovery.

Making experimentation easy is only half the equation. It’s just as important for leaders to make it safe to try new things and occasionally fail. When Reid Hoffman cofounded LinkedIn, he gave his team members an unusual goal: He dared them to fail 15% of the time. “In order to move fast,” he told them, “I expect you’ll make some foot faults.” Similarly, Reed Hastings, the cofounder and chairman of Netflix, would often challenge his team members when too many of their shows became hits. The lack of failure didn’t mean they were getting it right, he felt. It meant they weren’t taking enough risks.