2019 Offensive Personnel Analysis
If the 2010s are going to be remembered for anything on the field, it will be for the rise and eventual domination of 11 personnel -- one running back, one tight end, and three wide receivers. At the beginning of the decade, it held a mere plurality, being used on 39.5% of all snaps. By 2013, it passed the 50% mark. In every year but one between 2011 and 2018, 11 personnel usage increased, hitting a peak of 64.2% of all offensive plays two years ago. The Rams set a record, using 11 personnel on 92.3% of their offensive snaps. The 11 personnel takeover was inevitable.
Two years ago, only one team stood against the oncoming tide of 11 dominance: Kyle Shanahan's 49ers were the only squad to use 11 personnel on fewer than half their plays. And that looked to be that; all the hot new playcallers in the league were using three-wide sets as their bread and butter. Shanahan's offense, a relic learned from his father in the distant past of the 1990s, was the exception to modern football.
And if you don't see the "but then, in 2019" turn here, you really have to bone up on your foreshadowing.
But then, in 2019, Shanahan's 49ers no longer stood alone. Be it out of strategic intent or injury-induced necessity, four more teams used 11 personnel on fewer than half their plays last season -- the Baltimore Ravens, Philadelphia Eagles, Arizona Cardinals, and Minnesota Vikings. That's a fairly impressive list of offenses, too:
- Greg Roman's Ravens led the league in offensive DVOA;
- Shanahan's 49ers and Kevin Stefanski (and Gary Kubiak's) Vikings were in the top 10;
- Kliff Kingsbury's Cardinals jumped from dead last offensively in 2018 to 13th last season;
- and Doug Pederson's Eagles fought their way through countless offensive injuries to still finish on the positive side of the ledger.
That's a very solid list of resumes. And it's not like they were just dipping their toes in the non-11 pool; the Vikings only used 11 personnel on 21.0% of their offensive plays, the lowest total we've recorded since Greg Roman's 2012 49ers.
But it wasn't just a couple of teams zigging where everyone else zagged. Twenty-one teams saw their 11 personnel percentage drop from 2018 to 2019. A dozen teams saw their 11 personnel percentage drop by at least five points. Eight teams -- a full quarter of the league -- saw their 11 personnel percentage drop by at least ten percentage points. Was that enough to dethrone 11 personnel, or at least put it under 50% usage?
Before we go any further, we should clarify a few things. This is personnel data, not formation data. When Tarik Cohen goes out wide, he's still counted as a running back. When Robert Woods lines up in the backfield, he still counts as a wide receiver. We're using the standard numerical system where the first digit is the number of backs and the second digit is the number of tight ends. We count formations with six or more offensive linemen separately, rather than counting the offensive linemen as tight ends. A formation with two backs, one tight end, one wideout, and six offensive linemen is marked as "621" and not "22."
|NFL Offensive Personnel Groupings|
|Personnel||2018 Pct||2019 Pct||Difference||2019 DVOA|
Note that because we don't track personnel grouping on penalties, those negative plays are all missing from this analysis, so the average offensive DVOA for this table is 4.2% rather than 0.0%.
11 personnel remains the dominant formation throughout the league, used on approximately two-thirds of all passing plays and just under 45% of all rushing plays; it's still the default to which everything else is compared. It's no surprise, either; it remains the most successful of the commonly used formations by a significant margin. But after usage dropped nearly six percentage points last season, 11 personnel is now at its lowest point since 2015. We saw a blip in 2017, with 11 personnel falling by just 1.1%, and that set off speculation as to whether we had reached an equilibrium or if it was just a one-year dip. This drop is five times more significant than that one. Again, we can't be sure as to whether this will be a new trend or anything -- more data will be needed before we can make any sorts of declarations there -- but it's the most interesting development we've had to report on in quite some time.
Of course, even with this development, 11 personnel still made up the lion's share of all offensive snaps last season, so it makes sense to see how each team performed in and out of that split:
|DVOA in 11 Personnel|
I always find the lack of correlation between teams that use 11 personnel frequently and teams that use 11 personnel well to be interesting. This year, there was actually a negative correlation, as the teams that used 11 the least tended to be the best at it, at least in terms of raw DVOA -- there was no correlation between how often a team used 11 personnel and their relative strength, compared to their DVOA in other formations. Six of the top 10 offenses in the league were in the bottom 10 in 11 personnel frequency, with only the Cowboys really sticking out as a great 11 user. Seven of the worst 10 offenses in the league were among the 11 most frequent users of 11 personnel, with only the Broncos really sticking out as flailing in non-standard formations. I would hesitate before reading too much into those numbers after just one season -- suggesting that Cincinnati struggled on offense due to strategic reasons rather than an utter lack of healthy talent is a bold claim -- but for the most part, the talented offenses were finding ways to utilize second tight ends and fullbacks more than the struggling ones.
In fact, let's quantify that a little bit more. If we take 11 personnel as the default, there are a number of ways a team can swerve. They can bring a second tight end onto the field. They can pretend like it's 1973 and employ a fullback. They can bring on fourth and fifth receivers and just Air Raid the heck out of everyone. They can bring on a sixth offensive lineman and ground-and-pound. We can look at just how teams did -- and did not -- veer out of standard formations. Note that these categories do overlap somewhat; a 22 formation will count under both 2+ running backs and 2+ tight ends. Still, it's a useful measure of just how teams were using their personnel.
|Number of Plays in Different Personnel Groups|
Alright, now we can make some more specific observations. To wit:
Holy cow, Kliff Kingsbury. The Cardinals used more four-wide sets than the next five teams combined. They weren't particularly good at them, mind you -- a DVOA of just 0.9%, compared to a 9.2% DVOA in more standard formations -- but that's a personnel grouping that makes a statement. Specifically, that statement is "oh geez, my tight end is Maxx Williams?" Arizona only ran 641 snaps with a tight end on the field; everyone else had at least 830. That's a direct result of Kingsbury bringing his Air Raid offense to the big leagues in a way we've rarely seen before. Pittsburgh led all teams in 2018 with 97 snaps with four or more receivers, and that seemed like a lot way back then. Kingsbury doesn't need tight ends to run his offense; the Cardinals didn't draft any or pick any up in free agency. I suppose ideally DeAndre Hopkins is a slight improvement over Damiere Byrd in those four-wide sets. Get your cornerbacks ready, NFC West; you're going to need them. That heavy use of four-wide formations sets Arizona apart from the other four teams that went away from 11 personnel; they added beef, while Arizona was more looking for speed.
The 49ers retained their title as the team to use at least two running backs more than any other -- as you'd hope they would, considering how much they're paying Kyle Juszczyk. The only reason they didn't hit the 400-snap mark once again was Juszcyzk's knee injury that cost him a month of the season. No team in football relies on their fullback more than Shanahan's 49ers -- and with great results. The 49ers' 29.9% DVOA with two more backs led the league among teams with at least 50 such snaps; without a second back, that DVOA dropped to a dead-even 0.0%. The 49ers ran 296 plays in some variation of the I-formation, the most in the league; Minnesota and New England were the only other offenses with 200-plus snaps. The 49ers and Vikings rolled with fullbacks Juszcyzk and C.J. Ham, respectively, and were both significantly better with them on the field than not, bringing a tear to the eyes of old running back coaches. The Patriots didn't have the same luck, comparatively struggling out of two-back sets due to midseason injuries to James Develin and Jakob Johnson -- they used two backs on 173 snaps over the first six weeks but just 111 times after that as they were forced to adjust. In a world where most teams don't even carry a starting fullback, finding backups is a tricky business. (The Patriots ended up using a lot of linebacker Elandon Roberts.)
To be recorded as an I-formation, however, your quarterback has to be under center, and that is not the only way to get two backs onto the field at the same time. Baltimore had over 300 snaps with two backs in the backfield, but only 20 were listed as I-formation as the Ravens basically lived in the pistol. There were 964 plays last season charted as being in the pistol; Baltimore had 552 of them. 343 of those plays had at least two running backs on the field; Baltimore had 230 of those. Like Arizona, Baltimore was basically running a college offense, albeit one Greg Roman had used with success in San Francisco before. Baltimore's DVOA out of the pistol was actually worse than their DVOA in other formations (26.1% to 38.9%), but that's mostly due to the fact that Baltimore ran out of the pistol roughly three quarters of the time, with both running backs and Lamar Jackson being significant threats with their legs. Their passing DVOA dropped from 56.0% to 47.0% in the pistol, while their run DVOA jumped from 5.7% to 19.0%; even for the league's most efficient rushing team, passing still was better on average. Baltimore's offense is incredibly fun to watch, and other teams should copy it as soon as possible -- as soon as they find a dual-threat player like Jackson to plug in. Shouldn't be too difficult, right?
Minnesota is really interesting. Not only was 11 personnel not their main formation, it wasn't even their secondary formation -- the Vikings ran over 200 snaps in both 12 and 21 personnel groupings, with 101 more in 22 just for kicks, joining Baltimore as the only teams with 100 snaps in each of the three heftier formations. They joined the Ravens as the only team with four different personnel groupings each over 100 snaps. They had the single most radical change in personnel philosophy last season, dropping from using 11 personnel 67.3% of the time to just 21.0% of the time. This focus on tight formations with bigger bodies, especially on early downs, was a big part of their offseason offensive makeover. And did it work? Erm, well…
|Vikings Personnel Usage, 2019|
|Personnel||Snaps||DVOA||Pass Snaps||Pass DVOA||Run Snaps||Run DVOA|
While you don't want to quibble too much about a percentage point here or there, the Vikings were pretty clearly better when they had three wideouts on the field than they were in those bulkier formations. In their defense, they had injury concerns to deal with at the wideout position at points, but still. Kevin Stefanski's play calling was not ideal last season -- we devoted a decent chunk of Minnesota's chapter in Football Outsiders Almanac 2020 (now on sale!) talking about how their devotion to running on first down did not work, and here's another example of the team succeeding despite the decisions from the offensive staff.
Some more tidbits from our personnel files:
- The Saints squeaked out the DVOA lead over the Ravens with two or more tight ends on the field, 38.22% to 38.20%. Taysom Hill is charted based on where he lines up in the formation, so some of those may be formations with Hill as a tight end.
- The Seahawks continue to love their six-lineman sets, with their snaps with an extra lineman going from 199 two years ago to 222 last season. Their percentage of the league's six-lineman sets dropped, however, as more and more teams experimented with bringing in an extra lineman from time to time. Only the Seahawks and Saints were particularly effective there, though, each putting up an offensive DVOA above 10% with an extra blocker up front; the league average was -4.8%. With George Fant gone, it will be interesting to see if the Seahawks keep their six-lineman tendencies up, or if they just use an extra tight end to help block. Meanwhile, Saints fans shall applaud when Will Clapp enters the lineup, if he continues to perform well in New Orleans' jumbo packages.
- Not every team got the memo about 11 personnel becoming less cool in 2019. Both New York teams saw their usage jump by at least 10 percentage points. For the Giants, that was at least in part due to Mike Shula working up an offense in which Daniel Jones would feel more comfortable. For the Jets, Adam Gase has always been near the top of the league in 11 percentage, and at least some of their horrible results can be tied into Sam Darnold's mononucleosis. Exclude the three weeks Darnold missed and the Jets' DVOA in 11 personnel jumps to -13.4%, which is still terrible but no longer last-in-the-league terrible.
- Last year, we focused a lot on the Rams' lack of personnel diversity, as they used 11 personnel 97.5% of the time when Todd Gurley was healthy and effective. Gurley was neither healthy nor effective in 2019, so while they still nearly led the league in 11 personnel usage, it was down to a saner level. The missing plays were filled in by 12 personnel, primarily in short-yardage rushing situations.
- The Eagles were the only team to have a tight end on the field for every single snap. That will happen when you both have Zach Ertz and a M*A*S*H unit for a receiving corps.
- The most common personnel grouping to be "missing" was 21, with the Jaguars and Rams never trotting it out. The Jaguars did run one snap of 20 against the Titans, with Tyler Ervin stepping in for an injured Chris Conley in what would normally have been a four-wide set. As for the Rams, they've run exactly one play with more than one running back on the field since Sean McVay arrived in 2017 and, frankly, I'm not entirely sure that's not a transcription error with Derek Carrier listed as a running back or something.
31 comments, Last at 22 Jul 2020, 3:50pm
#22 by theTDC // Jul 21, 2020 - 6:26pm
I always knew that the McVay "11 personnel obsession," was simply him getting the most out of the talent on that Rams 2018 roster. That team had Cooks, Woods, and Kupp, along with the last year of healthy Gurley, and that was it. It doesn't take a genius to understand that 11 personnel simply puts your best talent on the field. I admired how eager he was to exploit this as opposed to artificially mixing it up, but that' all that was going on. This past year, injuries, and the emergence of Higbee at TE combined to make it extremely difficult to not add in that extra TE here or there, so the numbers came down.
I think the lesson the rest of the league should have taken was "get three great WR's," as opposed to "11 personnel is god."
#23 by Bryan Knowles // Jul 21, 2020 - 6:56pm
I mean, they were still second in 11 personnel usage, behind only one of his disciples, so it's still fair to say that McVay is pushing that 11 personnel agenda -- but 97.5% was not something that was going to be sustainable, and Gurley's lack of success last season forced the Rams to look elsewhere for production. It's the sign of a good coach to be able to adapt to circumstances like that, but I still feel that, if he had everything in place, he'd just trot out three receivers every play and call it a day.
#14 by Will Allen // Jul 21, 2020 - 7:51am
Look, no matter what rb, te, and wr personnel you place on the field, it's hard to run offense if the 5 bigs don't block. Zimmer got frustrated in 2018 not becauae he thought running plays were sure winners, but because he thought too mamy passing were going to get his not terribly mobile qb killed. How many times was Cousins hit in 2019, compared to 2018?
#17 by Aaron Brooks G… // Jul 21, 2020 - 9:00am
His sacks depends on source -- one site says 36, another 28. Both agree he had 40 in 2018.
However his hurries and QB hits changed a ton.
2018 was 40 sacks, another 39 QB hits (hit while throwing), and 97 hurries.
2019 was 28 sacks, 30 hits, and 48 hurries.
Cousins saw a ton less pressure in 2019, and his passing stats jump accordingly.
#18 by dank067 // Jul 21, 2020 - 10:54am
I mean, Cousins was sacked and hit at roughly the same rate adjusted for the number of dropbacks. It's good that the hurries went down, but overall it didn't amount to a tremendous year-to-year difference in pressure rate: 26.5% to 22.0% from 2018 to 2019.
For whatever reason PFR doesn't show pressure rate on their advanced passing season leaderboards, but if you search some QB stat pages you can see that neither of Cousins' pressure rates are especially high (that's obviously a function of how Cousins plays, too), and also that season-to-season swings of 4% are pretty common.
I certainly won't argue that it's better for QBs to absorb a higher volume of hits, but you don't have to commit to more running plays to reduce pressure rate. Dallas opened up their passing game in 2019, Dak Prescott had a career high number of dropbacks, and his pressure rate fell from 29.1% in 2018 to 18.8%. I think the Vikings were also running a good scheme and that they didn't need to limit Cousins' pass attempts just to realize those gains in efficiency, even with a crummy OL.
#20 by Will Allen // Jul 21, 2020 - 12:25pm
Opening up the passing game with the Cowboys pass blocking is vastly different than doing so with the Vikings pass blocking.
The more I think about it, the less sense it makes to seperate ball handling personnel groupings from the context of the bigs who are blocking for those groupings. In other words, causality is simply incredibly difficult to establish in football, and leads to all manner misconceptions, even by people who have rejected lazy conventional wisdom.
#25 by Will Allen // Jul 21, 2020 - 9:46pm
I should note that when the objective is avoiding having your 90 million dollar investment become a complete debacle, pressure rate on the 90 million dollar investment is a lesser concern than total number of severe hits on the 90 million dollar investment. This still remains fundamentally a game where the primary object is to more efficiently apply severe violence upon the opponent, and allow that violence to have a cumulative effect. Sometimes that gets lost in the analysis. Zimmer's been well aware for several years that his offensive line sucks at violence, and the implications of that fact. He's not some unthinking cro-magnon, pining for the days of Amos Alonzo Stagg.
#5 by theslothook // Jul 20, 2020 - 3:01pm
I'm glad Stefanski was mentioned here. I thought the Vikings offensive approach was baffling given their roster construction and effectiveness. I maintain it nearly cost them their playoff game against the Saints. Low variance game plans usually get you beat against superior opposition.
I'm perplexed to know what the browns saw that made them want to hire stefanski
#6 by Bryan Knowles // Jul 20, 2020 - 3:08pm
I don't think I'd call Stefanski's game planning or playcalling bad in and of itself, per se -- it just didn't really match-up with what the Vikings were good at or the personnel they had in place.
I'm slightly optimistic about Kubiak taking over, even after the Diggs trade.
#7 by theslothook // Jul 20, 2020 - 3:32pm
I guess I have a bigger issue with it.
His approach in those playoff games reflected conservative/hide the qb offense. To me, you only go that route because a) your qb is horrible and likely to lose you the game with turnovers or b) your defense is awesome and will keep the game low scoring.
Otherwise, I feel like that style just sets you up to get upset or blown out when you play that way. It might have worked in the early 2000s but its not going to work today.
#11 by dank067 // Jul 20, 2020 - 7:22pm
The Vikings approach last season can probably be explained better as "hide the OL" than "hide the QB." I would still agree that it featured a lot of outdated thinking, but i) they were coming off a season where Zimmer publicly lambasted and then fired their previous O coordinator for not running the ball enough, so some of the playcalling emphasis was clearly coming from above Stefanski, and ii) they did manage to implement a pretty explosive play action passing game.
Stefanski has worked with a diverse array of offensive coaches during his career, so I'm interested to see what he's going to do in Cleveland now that he's running the show with a lot of weapons at his disposal. Not a great-looking OL there again though.
#12 by theslothook // Jul 20, 2020 - 8:18pm
The problem with the line of thinking listed above(and I guess this is a shot at Zimmer) is running that style may exacerbate your situation. If runs on 1st down repeatedly yields little yardage, then you are asking your o line to pass block long enough to ensure a throw makes it to the sticks. Essentially sacrificing first down so that your third down becomes more onerous. Furthermore, there is no correlation between play action and run success/run propensity. They can have a wonderful play action game without being wedded to the philosophy of run first.
#13 by dank067 // Jul 20, 2020 - 8:34pm
I'm with you 100% on why going run heavy on 1st down can end up making life more difficult for both QBs and offensive lines. I think the Vikings offense last year ultimately had a nice scheme but didn't quite live up to its potential thanks to sub-optimal play sequencing and personnel usage (keeping in mind that Thielen was hurt/had a down year and they didn't have much else at receiver - could be a problem this year too).
I don't know if that will change for MIN this year with Kubiak stepping into the OC role, but like I mentioned I am interested to see what Stefanski does now that he's on his own.
#9 by theslothook // Jul 20, 2020 - 3:50pm
This is probably a contrarian take but I would not look to hire Eric Bieniemy. Not to say he couldn't be a good head coach, but its very much the laziest most expected thing to do; hiring the coordinator of a successful offense or defense. this is especially true when we consider who the head coach of the chiefs is.
That by the way is in contrast to coordinators who have run successful offenses for a variety of head coaches. I'm very surprised Greg Roman did not get considered for head coaching. Or Dave Toub.
#27 by Will Allen // Jul 22, 2020 - 9:42am
Pat Shurmur is a good h.c., I suspect, who had bad luck in the jobs he's been offered. The work he did in Minnesota as o.c., in the context of a historically bad injury tsunami, including getting his injury prone starting qb in week 1, was off the charts good.
#3 by Bryan Knowles // Jul 20, 2020 - 1:47pm
Ignoring snap count minimums, the best personnel grouping in football last season was 32, with a DVOA of 148.9% on three snaps, all by Green Bay, all resulting in touchdowns from Jamaal Williams. Weirdly for a formation with three running backs, one of them was even listed as empty back spread, with Williams, Aaron Jones, Danny Vitale, Jimmy Graham and Marcedes Lewis all split out various degrees of wide; none of them playing in the backfield OR on tight on the line. Weird, weird stuff.
Increasing our snap counts from there, the most effective formations become:
*721 (54.3% DVOA, 8 snaps), mostly run by the Chargers at the goal line as a Melvin Gordon showcase
*03 (40.8% DVOA, 15 snaps). Detroit's the one that rocked that the most, though only when Matt Stafford was healthy; they had a zillion tight ends to manage.
*23 (18.5% DVOA, 69 snaps), a formation the Ravens and Bills liked in the red zone, and one of the few situations when Lamar Jackson WOULD get under center.
*610 (10.8% DVOA, 90 snaps), of which 60 snaps were Seattle and their George Fant: Secret Tight End package
*10 (9.0% DVOA, 780 snaps)
*22 (6.3% DVOA, 906 snaps)
*11 (6.0% DVOA, 18,904 snaps)
Going the other way, the worst formation in football was 711, resulting in zero successes on seven tries by Pittsburgh and Seattle, for a total DVOA of -118.1%. They were all Rashad Penny or Benny Snell runs, resulting in a combined total of zero yards, with zero successful plays. Both Seattle and Pittsburgh busted the formation out in exactly one game due to injury constraints, and never touched it again.
Increasing our snap counts from there, the least effective formations become:
*00 (-67.9% DVOA, 23 snaps). The Steelers ran this formation 14 times against New England in Week 1 with almost no success; they only went back to it three times after that.
*620 (-26.4% DVOA, 26 snaps); Washington was the most keen on experimenting with this, though somehow it resulted in two sacks on Dwayne Haskins despite the extra lineman.
*621 (-10.2% DVOA, 143 snaps)
*20 (-10.2% DVOA, 340 snaps)
*612 (-4.2% DVOA, 438 snaps)
*611 (-2.9% DVOA, 505 snaps)
*13 (-1.4% DVOA, 886 snaps)
*12 (1.8% DVOA, 6,201 snaps)
*11 (6.0% DVOA, 18,904 snaps)
#1 by Aaron Brooks G… // Jul 20, 2020 - 1:33pm
I swear, if I were Detroit, I'd go all in and play a 600.
You can't block, can't run, and your TEs suck. Your only hope is Stafford finding someone open, and you only have four decent receivers. So just go all in and replace your TEs and WRs with receivers and keep a 6th blocker in to maybe trip someone as they fall down.
#30 by DraftMan // Jul 22, 2020 - 3:11pm
Then why not go even further on the "all-in" plan like Jon Bois, who's out out here trotting out 10 personnel but having them line up in a 31 formation to have NINE blockers?