As he reviewed his career this offseason in search of areas to improve, one play inspired Kirk Cousins’s curiosity. He recalled a quarterback sneak that had keyed a Washington Redskins victory over the Green Bay Packers. If a sneak had been that important, he figured, he wanted to ensure he could convert another if needed. He asked members of the Minnesota Vikings’ analytics staff to tell him his success rate on quarterback sneaks.

“Then I said after that conversation, ‘Hey, this is good stuff,’ ” Cousins said. “ ‘If ever you have information that arises, please send it my way. Don’t be a stranger. I might not be smart enough to ask the question, but can you come to me?’ ”

Shortly thereafter, a Vikings analytics staffer provided Cousins information that helps explain an overlooked NFL trend. He showed Cousins his passing statistics with and without play-action. On straight dropbacks, Cousins’s numbers resembled those of a mid-tier NFL starter. When he faked a running play before throwing, Cousins looked like a superstar on paper.

Cousins, whose recent hot streak has helped make the Vikings into a playoff contender, is an extreme example of the difference between using play-action and not, but he is also representative. Almost uniformly, NFL quarterbacks succeed at a higher rate when they use play-action fakes than when they don’t. It’s clear that coaches and players are catching on, and given the increased use of data within the sport, we may only be seeing the start of an information-fueled play-action boom.

The value of play-action is easy to understand. A team can either disguise its intentions or not. And yet dogmatic thinking has limited its use more than necessary. The play-action pass may be turning into the NFL’s version of the three-point shot in the NBA: It is a simple, ancient tactic that delivers an obvious strategic advantage, but for generations it went underutilized in a way that seems irrational, in retrospect.

Through seven weeks, according to data mined by Pro Football Focus, NFL quarterbacks have punched up an 88.3 quarterback rating and thrown for 6.93 yards per attempt when not using play-action. When employing a run fake, their quarterback rating is 104.8 while gaining 8.93 yards per attempt. In terms of black-and-white numbers, play-action turns the average quarterback from rookie Kyler Murray to superstar Aaron Rodgers.

This year’s results are not a fluke. Last year, NFL quarterbacks not using play-action averaged seven yards per attempt with a 90.0 rating. With play-action, those numbers jumped to 8.53 and 102.1.

Some of that difference is noise — teams are far less likely to use play-action passes on third-and-long, plays with inherently less chance of success. But even when early-down plays are separated out, the efficacy of play-action shines through. According to Warren Sharp, who owns the analytical company Sharp Football and consults with NFL teams, teams average 8.5 yards on early-down play-action passes and 7.5 on straight dropbacks — a significant difference.

NFL teams are taking notice. Over the past three years, per PFF’s data, the percentage of passes using play-action has crept from 21.4 to 24.1 to 24.7. In 2016, playing under then-Atlanta Falcons offensive coordinator Kyle Shanahan, Matt Ryan led the NFL by using play-action on 27.6 percent of his dropbacks. This season, 10 quarterbacks have deployed play-action with greater frequency, and five have used play-action on at least a third of their dropbacks.

No quarterback, perhaps, has benefited more from increased play-action than Cousins. The Vikings have used it on 35.7 percent of Cousins’s dropbacks, second in the NFL to the Indianapolis Colts’ Jacoby Brissett. On those plays, Cousins has completed 71.1 percent of his passes at 11.1 yards per attempt, posting an astronomical 140.2 quarterback rating (out of a possible 158.3).

It has helped fuel Cousins’s recent three-game resurgence after a rocky start to the season that had fans and even some of his own teammates critical of his performance. Minnesota is now 5-2 entering Thursday night’s home game vs. the Redskins, and Cousins leads the league in passer rating, narrowly edging MVP candidates Russell Wilson and Patrick Mahomes.

Cousins, 31, became convinced this summer, after Minnesota’s analytics department showed him the data. Cousins ran play-action on 20.8 percent of his attempts last year, only 17th in the NFL, but when he did he completed 77.1 percent of his passes, best in the NFL, for an excellent 116.1 rating.

“The main thing was, even if you’re not running the ball effectively, still use play-action,” Cousins said during a training camp interview this summer. “It’s still going to slow the pass rush down, make linebackers feel unsettled as to whether to get in a zone drop or go upfield and fit their gap. Usually the routes are designed for bigger plays, and so you’re able to get bigger plays. It basically said, never get away from it. The numbers would say just keep going back to the well.”

Cousins sounded like an acolyte of the burgeoning analytics community, which has debunked myths associated with play-action passes. In 2018, Ben Baldwin of Football Outsiders wrote a convincing article arguing that setting up play-action passes using running plays is unnecessary. Sharp said player-tracking data shows linebackers pause or creep toward the line on play-fakes regardless of how many play-action passes an offense has used or what running back is in the backfield. (In Minnesota’s case, Dalvin Cook leads the NFL in rushing through seven weeks.)

“Whether it’s play-action attempt 10, 12, 17 in a game, you’re still luring in those defenders and creating the little extra edges in that passing attack based upon spacing,” Sharp said. “And that’s the whole purpose of a play fake. You just need to keep [a defender] on his heels for a split-second. The defense is hard-wired mentally to react to play-action.”

It probably shouldn’t have taken in-depth studies to realize faking a run before passing is more effective than not using deception, much like it shouldn’t have taken advanced metrics to realize a three-pointer in basketball was more efficient than most two-point shots. But those analytics correlate to the recent bump in play-action passes. “We’re arming them with this information they might not have recognized,” Sharp said.

This summer, Vikings offensive coordinator Kevin Stefanski smiled when asked whether he believed play-action should be used more often across the league. “I’ve read the articles,” he said. “I mean, sure, there’s data to support that. We’re well aware of all of it.”

Play-action passes also carry an inherent design benefit. The fake slows the pass rush, and the time it takes to execute the fake allows more time for receivers to run slow-developing routes, which tend to create the best chances for explosive gains. And the deeper dropbacks allow quarterbacks more space and time to read the field.

“I hated running double-move routes on a straight dropback,” former NFL quarterback Doug Flutie said. “I dropped back, set, give a little pump fake. And as I reload, my offensive line is falling in my face, or guys are getting pressure, and the vision becomes more clouded. With a hard play-action, it gives you something to do while the move is going on. You’re getting deeper, and you’re ready to come off it, and you’re ready to throw as that guy’s made that double move.”

The NFL has begun to act on the realization that play-action passes are an underused threat. As teams continue to evolve, they may hit a point at which the effectiveness diminishes. That point does not seem to be close.

“Across the board, on average, I still think we’ve never reached the point where there’s too much play-action such that we don’t see a benefit to teams,” Sharp said. “Whether or not that point exists remains to be seen.”

Read more on the NFL: