Disappointment stretched across Josh Allen’s face as he recounted his final play of the season.
“They gave a good look,” Allen said during his postgame press conference a week ago Sunday, describing Kansas City’s disguised zone pressure on 4th-and-5. During Allen’s dummy cadence, he explained, the Chiefs completely concealed corner Trent McDuffie’s blitz plans. The Buffalo Bills offensive line committed to sliding the other way.
Then the snap … but there Allen’s retelling trailed off.
Never mind his 10 yards of frantic scrambling, his 35-yard fluttering heave, the football slipping through tight end Dalton Kincaid’s outstretched arms. Buffalo’s latest hunt for its first Super Bowl victory might as well have ended before Allen took a step.
“Yeah … ,” he said.
NFL plays no longer start with the snap. Offenses send players in motion six times as frequently as they used to, hunting for a head start or a tip-off. Defenses mask their true intentions as best they can. Signal-callers on both sides play cat-and-mouse with audibles and adjustments, deceptions flying back and forth. Player tracking data has only heightened the importance of those moments, indicating both how much value slight shifts can create as well as how much a minor move might betray what’s to come. Artificial intelligence threatens to ratchet up the complexity even more.
Watching from afar, analysts, broadcasters and fans are left to keep up.
ESPN’s NFL video tracking team began measuring pre-snap motion in 2019, and ESPN analytics writer Seth Walder dug into the new data midway through that year. “I remember where I was,” Walder said of first reviewing the information. “We were seeing offenses were more efficient when they had a man in motion at the snap.”
Since then, the use of motion-at-the-snap has exploded across the league. ESPN’s measurements now go back to 2017, when no team was running players at the snap as much as 10% of the time. In 2024, offenses did so 25% of the time on average. The Miami Dolphins led the league, using motion-at-the-snap on 56% of plays.
“This is a dramatic shift in the NFL,” Walder said.
While motion can help identify whether a defense is playing man (if a single defender trails the moving offensive player) or zone (if the defense only shifts slightly), Walder sees other benefits. Movement just before the snap can create confusion among defenders whose responsibilities—often based on minute differences in offensive formations—have suddenly changed. Plus, speedy weapons are now already in second gear.
The downside? Motion adds complexity that can trigger miscues. It costs time. It can tip off savvy defenses. And it can make last-minute adjustments more difficult; after all, the offense’s plans are already … in motion.
So if you have to take your eyes off Super Bowl LIX at any point next Sunday, don’t do it during the precious seconds that often augur explosive plays.
The Chiefs used motion in several key instances during their 2023 Super Bowl victory over Philadelphia, most notably on a pair of fake jet swing passes that were called after KC coaches picked up on a deficiency in the way Eagles defenders responded to the activity. In the week leading up to the game, the Chiefs specifically practiced their pre-snap routes.
This year, the Eagles jumped from dead last in motion-at-the-snap to roughly league average, Walder said. The Bills, for what it’s worth, sent wideout Khalil Shakir in pre-snap motion on their failed fourth down at the end of the conference title game, though Allen never had time to look his way.
The 2025 Big Data Bowl, a $100,000 crowdsourced data science competition co-run by the NFL, asked football geeks to study “pre-snap behavior to predict and better understand NFL team and player tendencies.”
“The reality about this specific prompt: Teams were asking for this years ago,” NFL senior director of football data and analytics Mike Lopez said. “The stuff that happens before the snap has become way more important in the last couple years.”
Data science competition finalists developed tools that predict defensive schemes based on player alignments, identify whether a tight end is likely to block or not on a play, score defenses’ ability to disguise coverages and more.
Following the league’s 2011 CBA, TV production crews have had the benefit of microphones attached to offensive linemen, bringing fans closer to the pre-snap action. That access has been critical, because producers are now forced to cut away from replays faster to capture more of that line-of-scrimmage chicanery.
As QBs use every second to analyze defenses and prepare their assaults, 28% of plays in 2024 began with less than five seconds on the play clock, the most since Next Gen Stats tracking started in 2016. Partially as a result, 2024 games saw an average of 124.4 plays, the second-fewest since 1992, according to Pro Football Reference data.
Some broadcasts are also turning to higher camera angles to capture the full complexity of wide receiver and defensive back movement, both before and after the snap.
“We really want to show people the game from the quarterback’s eyes,” Thursday Night Football analytics expert Sam Schwartzstein said. Amazon’s “Prime Vision with Next Gen Stats” feed uses artificial intelligence models to highlight potential blitzers and possible open areas, replicating aspects of a QB’s pre-snap process.
While offenses have learned to use every second, defenses are also taking advantage of the extra time, primarily through duplicitous presentations and audibles in response to offensive calls. In particular, Schwartzstein has noticed multiple instances where defenses respond to motion as if they’re playing man-to-man coverage, only to drop into zone.
“It’s iterative; now defenses can muddy their own looks up once they figured out what you’re trying to do,” Schwartzstein said. “That’s why football will never be a solved sport.”
It’s not just offenses taking longer before each play, Schwartzstein added, but defenses forcing them to.
It seems hardly coincidental that the focus on pre-snap looks has exploded exactly as our ability to analyze those moments has improved. “What gets measured, gets managed,” the saying goes. And player location data, along with advanced analytical techniques, have made it much easier for coaches to evaluate the impact of subtle tweaks over large sample sizes.
New generative functions present even more predictive capabilities. The Chiefs identified the Eagles’ vulnerability to reverse motion passes in 2023 after seeing the Jaguars run a similar play against Philly earlier that season. How many other holes could AI find, when scouring data from every single play?
Even more inputs could be coming. One Big Data Bowl paper used QB positioning to identify when teams audible and track how successful they were with their backup plays. In theory, audio analysis could be added to better translate what happens after a team breaks the huddle. Skeleton-level tracking would give teams even more ways to look for potential edges and indicators.
"I always tell people, 'I have the biggest smile on my face when I'm blitzing,'" McDuffie, the Chiefs corner, said last week. “That's my dead giveaway if you line up against me.”
If only Josh Allen could have known.
