The most visible tactical change in the sport over the past decade is how often teams attempt fourth down conversions, and it came from analysis rather than from courage.
The old default
Punt on fourth down outside field goal range, kick inside it, almost regardless of distance.
Which was conventional to the point of being unquestioned.
Deviating from it drew criticism whenever it failed and produced little credit when it worked.
What the analysis showed
Expected points models, built from historical outcomes by field position and down, indicated that punting in many situations gave away value.
Which was published academically well before it was adopted.
The gap between what the numbers suggested and what coaches did persisted for years.
Why the gap existed
Asymmetric criticism.
A failed conversion is visible and attributable, while a punt that quietly cedes expected value is invisible.
Which is a straightforward incentive problem rather than a failure of intelligence, and it explains the delay better than anything else.
What changed it
Analytics staff embedded in organisations, public models that made the reasoning visible, and a handful of coaches willing to accept the criticism.
Which normalised it once results did not obviously suffer.
Once several successful teams did it, the reputational cost fell for everyone else.
The nuances the models capture
Distance to go, field position, score, time remaining and opponent quality all shift the recommendation.
Which means the answer is genuinely situational rather than a blanket policy of aggression.
Late in games, win probability rather than expected points becomes the correct objective, and the two can disagree.
What models do not capture well
The specific personnel available, injuries, weather and the state of a particular defence.
Which is why a model output is an input to a decision rather than the decision itself.
Coaches who treat it as either gospel or noise are both mishandling it.
The knock-on effects
Punters and kickers see different usage, defensive strategy on third down has adjusted, and play-calling on third-and-medium changed because a fourth down attempt is now assumed.
Which is what a genuine tactical shift looks like — it changes the decisions around it, not just the one it applies to.
Win probability models
Estimate the chance of victory from any game state, built from historical data.
Which is what late-game decisions should optimise rather than expected points.
The two frameworks agree in the middle of games and diverge sharply near the end.
Two-point conversions
The same analytical shift changed conversion decisions.
Which now follow charts derived from score differentials rather than convention.
Going for two earlier when trailing by specific margins is now standard and was once rare.
Kickoff and onside rules
Rule changes reducing onside kick success altered late-game strategy considerably.
Which made trailing comebacks harder and shifted when teams begin taking risks.
Alternative proposals to replace the onside kick have been trialled in various forms.
Fourth down and coaching security
Coaches with secure positions take more analytically supported risks than those under pressure.
Which is observable and is entirely rational given how failures are received.
Public availability
Win probability and fourth down models are now published live during broadcasts.
Which has changed how decisions are discussed and holds coaches to a visible standard.
Field goal range expansion
Kicking accuracy at distance has improved substantially over decades.
Which changed the calculation by extending the range where a kick is a realistic alternative.
It also shortened the field where going for it is clearly preferable.
Situational exceptions
Weather, a struggling offence or an exhausted defence can override a model's recommendation.
Which is where coaching judgement legitimately enters.
The best decisions combine the baseline with information the model does not have.
College and lower levels
Adoption has spread downward, with high school programmes now using similar frameworks.
Which has changed how the game is coached at development levels.
What this illustrates generally
A demonstrably better decision took years to adopt because of how failure was punished.
Which is a pattern visible well beyond sport.
The lesson is about incentives rather than about football.
Watching for it
Note the down, distance and score, and the decision usually becomes predictable.
Roster consequences
Teams now build offensive personnel with short-yardage conversion in mind.
Which affects who gets a roster spot at the margins.
The value of a reliable one-yard conversion has risen measurably.
Where it stands
Aggression has increased league-wide and remains below what most models recommend, which is an ongoing gap.
Broadcast presentation
Live probability graphics have educated audiences and raised expectations of coaches.
Which has changed the tone of criticism from instinct to arithmetic.
Whether that improves the discussion is debatable, and it has certainly changed it.
Closing
A better decision existed for years and was not taken, and understanding why is more interesting than the decision itself.
The same reasoning is now applied to timeout usage, clock management and end-of-half decisions, all of which have shifted measurably.