Words are one thing, but action is another. Many fans would love to have the ability to watch every NFL game each and every week, but only about five time slots are used for the 16 or so games each week, forcing most fans to choose which games to devote their attention to. By comparing viewership data for individual games between Weeks 10 and 17 in 2020, we can look at which factors fans care about the most. The old adage is that offense sells tickets (and attracts viewers), but is that born out by the numbers in 2020?
To evaluate this idea, we chose to eliminate the playoffs and all games before Week 10. The playoffs are a viewership anomaly because there are no overlapping games; reducing our data to the final eight weeks of the season allows us to use season-long statistics under the theory that fans will generally know which offenses/defenses are better, which teams they like, etc. Additional factors to consider are the network that a game is broadcast on and whether it’s a “split-ticket” game. Some networks are more entrenched and may attract more viewers, even to the point of hosting “split-tickets,” where they give different games to different sections of the country while simply reporting a single viewership number. We can normalize for the hosting network by converting reported viewership into a percentage above (or below) the sample average of that network. For the split-tickets, the reported viewership was split based on the percentage of the country that the game was broadcast to. With the data set up and 34 team variables collected (averaged between the two competing teams for most variables), individual linear regressions between the viewership percentage over expected and each variable can describe which factors were truly most important.
With this graph, we’ve combined each variable into groups in order to get the general idea of which kinds of variables are most predictive of a high television audience. The longer bars indicate a stronger relationship with the reported viewership, so the preseason Super Bowl odds were almost entirely unrelated to TV viewers over expected while Overall Offense was highly related. Let’s break down some of the more interesting categories.
Overall Offense, Passing, and QBs
Perhaps unsurprisingly, these three categories take a commanding lead in this study. Of all 34 individual variables, All-Purpose Yards was the second-most related to television viewership over expected, followed closely by total offensive expected points added (EPA), passing offense, and both scoring/total offense. The style of offense was also important here, as viewers strongly preferred a passing-oriented attack to the run game and also preferred a team that could push the ball downfield with a high net yards per attempt. With teams like Green Bay, Tampa Bay, New Orleans and Kansas City dominating in the rankings, this comes as little surprise.
Quarterbacks are the focal point of every team both on and off the field, so we used additional factors to evaluate their impact. Past their draft position (which had a middling relationship to viewership with a 0.57 p-value), the combined number of career Pro Bowls was the single most related factor to viewership, best demonstrated by strong ratings of the Week 12 Super Bowl preview between the Buccaneers and Chiefs and the Colts-Packers match of Week 11. The combined number of Twitter followers for the two starting QBs was also highly relevant and the third-highest individual variable, with Seattle’s Russell Wilson both leading all quarterbacks in Twitter followers and being involved in the six-highest matchups of Twitter-famous passers.
The most surprising relationship near the top involved return yards. This has some quick explanations in that teams with more return yards tend to be seen as more exciting and may attract fans, but seeing a top spot taken by a factor that only includes special teams is astounding. With media-friendly teams like the Colts, Bills, Cowboys and Patriots near the top of the return leaderboards, it’s a bit more apparent as to how this result came to be. This is highlighted best by the most over-performing matchup in our sample, the Dallas-Washington Thanksgiving Day game that boasted the 13th-best combination of return games. The NFL has been trying to eliminate kickoffs, but fans clearly still love the play.
Wins and Penalties/Giveaways
Fans have always flocked to good teams, and continue to do so, but it is interesting how strong passing offenses and popular QBs have now taken precedence for fans over winning. Although matchups between teams with a higher combined win percentage still do well, like Week 15’s Saints-Chiefs game, these games don’t always produce more viewers than expected, with 70% of the 10 sampled games with the lowest combined win percentages actually producing above expected viewership. Perhaps fans simply don’t like mid-level teams, and would rather watch outstanding teams against poor teams? The data is similar, but indeed matchups between teams that finished further away from 0.500 actually were more related to viewership over expected than simple win percentage. However, the margin of victory was not nearly as related, with a split between the fans who preferred to watch blowouts and those who wanted closer games.
Penalties, on the other hand, were absolutely a mood-killer. Fans were willing to stomach offensive penalties but had no such regard for defensive penalties, one of the strongest relationships in the sample. Giveaways were similarly disdained, even though the counterpart category of takeaways (categorized in overall defense) served little relation to viewership. This could be a corollary of better teams simply committing fewer defensive penalties, but the weaker teams in the top-ten of preventing defensive penalties, like Minnesota, still were capable of strong viewership when paired against better teams.
Start Time of Game, Franchise History, and Market Size
Three of the most-commonly referenced factors to a better viewership base each showed little relationship to viewers over expected. The most-overviewed games were spread out over each of the 1:00, 4:00, and 8:00 time blocks, and each slot averaged between minus-5% and 5% viewership over expected. Franchise history is difficult to put into numbers, but by evaluating the Super Bowl wins in team history we can get a rough idea of the more historic teams like San Francisco and Dallas. Although the aforementioned Cowboys game on Thanksgiving involved teams with a combined eight Super Bowl titles and drew fantastic ratings, this was an outlier; other games between Pittsburgh and Washington as well as New England and Baltimore each performed below expectations. Finally, large-market teams weren’t related to higher viewership, but this is likely more of an outlier year with the 13 largest market teams experiencing some level of a down year while nine of the 13 smallest markets going to the playoffs. With new QBs littering the large markets, this may return to form in 2021.
Defense, Coaching, and Preseason Odds
Preseason odds came in dead last of every potential factor, without any hint of a relationship to viewers over expected. This does not mean that worse odds equals more viewers, but rather that the data is almost entirely random. Coaching reputation, specifically in the number of years that the HC had been with the franchise and the number of Coach of the Year awards gained within that span, similarly had almost no relationship to viewers over expected. Somewhat surprisingly, overall defense came in third-to-last of the categories. While one defensive variable (defensive EPA) showed a modest relation to viewers over expected (0.50 p-value), total defense, scoring defense and takeaways each were almost entirely unrelated. In the 2020s, defense may not win championships or fans.
The NFL is a media behemoth that is one of the last pieces of traditional programming preventing a full cord-cutting revolution — which speaks to the way they’ve been able to properly identify, predict and encourage what exactly makes a fan engage with the product. By protecting passers and limiting defensive contact, the league has bet on scoring and passing as the primary driver of fan involvement. This was validated over the second half of 2020, as offensive prowess overtook winning itself as a factor that can drive fans to action, even if it’s just pushing a button on their TV remote. With the surprising result of return yardage showing a relationship to viewership over expected, the next challenge will be whether they can retain this excitement while protecting their players. At least where viewership is involved, the NFL will probably get it right.