Examining the Steelers’ doomed final play against the Broncos

Ugh.

That’s my gut reaction to what transpired in Denver on Sunday. After escaping Jacksonville with a win that probably should have been a loss, we exited Denver with a loss that should have been a win. No moment better captured the frustration that was Sunday’s effort than the ill-fated final offensive play, when Ben Roethlisberger threw a brutal game-ending interception with the Steelers poised to score and (likely) force overtime. How in the name of Cliff Stoudt did that happen? Here’s a breakdown.

The Situation

Denver 24, Pittsburgh 17. Steelers ball, 1st and goal on the Denver 3 yard line, 1:57 to play.

The Steelers come out on 1st down in an 11 personnel set (3 WR, 1 RB, 1 TE) with a pro look into the boundary and a slot look to the field. James Conner is in the backfield to the right of Roethlisberger. They fake sweep to Conner and run a fade to Juju Smith-Schuster out of the slot. Juju is slightly open but Ben overthrows him by a hair.

On 2nd down they are in the same 2x2 set and they run the same concept with one significant twist. Rather than run the fade, Juju, in the slot below, runs a slant while the offensive line again blocks sweep for Conner. This is likely an RPO, or run-pass option, whereby Roethlisberger reads an unblocked defender and decides to give the ball to Conner (run) or throw it into the area voided by the read key (pass). Here, Roethlisberger hands the ball off because his read key sits in the alley and does not pursue the sweep, thereby denying the slant.

These two plays are clearly packaged, giving Ben an RPO on each based on the look he was getting from the defense. It’s possible the Steelers saw Juju win the slant on 2nd down and decided to try it again on 3rd down from a different look.

3rd Down

The Steelers line up in a different formation. They bring Jesse James over to the slot side to give them Tight End-Trips to the field, with AB isolated to the boundary. Conner is still to Ben’s right.

They are going to run the same slant-RPO as they did on 2nd down except they’re going to change the receiver and the read key. Rather than having Juju run the slant and read the backer to the field, they are going to have AB run the slant and read the backer to the boundary. What they’re trying to do with this package is to RPO the backers while setting the Bronco DBs up to win the slant (Juju runs fade on 1st down to get his defender to widen then tries to beat him inside with the slant on 2nd; AB runs fade on 2nd down in order to beat his corner inside on 3rd). They flip James to the other side to remove a defender to give AB more room to run the slant and they change the run play from a sweep to a trap because without the tight end on that side they don’t have enough players to block the sweep. Otherwise, these are simply two different versions of the same basic concept.

And what of the concept? I find the design to be a wonderful example of modern football thinking. It displays creativity and a complex understanding of scheme. It gives the quarterback the ability to make decisions that avoid dead plays based on pre-snap alignment. It prescribes answers for just about anything a defense can throw at it. There’s just one problem: to make it work, it also requires precise ball-handling and perfect execution.

What Went Wrong

Plenty. First, and most crucially, the snap is low and wide. Much like the season opener in Cleveland, where a bad snap on an RPO resulted in a fumble that let the Browns tie the game late in the 4th quarter, the snap here disrupted the timing of the play. As seen below, Roethlisberger reaches down and to his left for the ball. By the time he gets his head up to ride Conner and read his key, Conner has run into him.

Rather than throw the ball away, or pull it and run it himself behind Conner (this is not technically an “option” here but it would be a smart improv, given how the play transpired), Roethlisberger tries to force the ball to Brown.

This presents a new set of problems. For starters, Brown is covered tightly. Even if Roethlisberger gets the ball off cleanly (which he doesn’t), the throw is going to be broken up or perhaps picked by the corner (in the frame below, the arrow shows how the corner has undercut Brown to get into position for the interception). Instead, because of how late the throw comes out, Shelby Harris, the nose tackle, slides off of the block of Maurkice Pouncey and makes the interception.

Watching live, I thought Harris baited Roethlisberger by acting as though he was rushing and then dropping into coverage. On replay, however, you see it is not a creative coverage scheme by the Broncos. It is simply a poor snap compounded by a bad decision from QB7.

58Steel was kind enough to lend his expertise and put the play together as a GIF. Here it is in complete form:

https://media.giphy.com/media/lcMkbgSwkJioOSgP7h/giphy.gif

To RPO or Not?

This is the second time now we’ve turned the ball over while running an RPO concept in the waning moments of a football game when ball security should have been a top priority. It leads to a natural question: should we use so many RPOs? And if so, when?

I’m a huge fan of RPOs. I believe they put defenses at a competitive disadvantage by using the thing that makes them most effective — aggression — against themselves. A defender is allowed to run unblocked to where he believes the ball is headed only to have the carpet pulled out from under him when the ball is delivered into the area he has vacated. If the defender gets wise and sits for the pass option, there is one less hat attacking the run and the offense has a numbers advantage at the point of attack. As a friend of mine often says about the concept, “That’s just cheating.”

Two elements are essential for an RPO to succeed, however. The first is a great snap. RPOs are built on timing. The quarterback must place the ball at the mesh point with the running back at precisely the right time in order to read his defensive key and make a decision on whether to give or pull the ball. If the mesh point is off, which it often is in the event of a bad snap, there is confusion between the RB and QB. Unlike a traditional handoff, where the QB places the ball into a pocket created by the RB, the RB must find the mesh on an RPO because the QB’s eyes are on his read key. Once he finds the ball, the RB is often instructed to clamp down on it when the QB removes his bottom hand from the mesh. If the snap is bad and the mesh is compromised, or the RB cannot determine whether the QB wants to give the ball or pull it, chaos ensues. Like it did against Cleveland. And again on Sunday.

The second necessary ingredient for a successful RPO is a quarterback who can make quick and accurate decisions. Ben Roethlisberger is certainly that quarterback. You don’t build a Hall of Fame resume if you can’t think on your feet. Unfortunately, Big Ben is also a gunslinger, the type of QB who believes he can make any play no matter the circumstance. Sometimes, as witnessed by his remarkable touchdown pass to Santonio Holmes to win Super Bowl 43, which was thrown into the narrowest possible window with the game on the line and was inches from being intercepted, it results in glory. Other times, like the fake spike play at the end of last year’s Patriots game, it gets the best of him. Yesterday it did the latter.

The smart play yesterday was for Roethlsiberger to realize the bad snap had compromised the play, to eat the ball and to play 4th down. There was plenty of time on the clock and the Steelers still had a good shot to score from inside the 5 yard line. There was no chance - zero chance - for Ben to complete that throw to AB given how things transpired. The snap threw off the timing, AB got too far towards the middle of the field on his route, the corner undercut the slant, Conner knocked Ben off balance so he couldn’t get anything on the throw. Everything about that scenario said Ben should have eaten the football or thrown it away. That’s not Big Ben, though. Never has been, probably never will. It’s what makes him great. Most of the time.

I wouldn’t give up on RPOs if I were Randy Fichtner. They give Roethlisberger the flexibility in the offense he was denied under Todd Haley and they present a myriad of options to make a defense wrong. The three plays we ran near the goal line on Sunday were brilliantly -designed. Sometimes you can out-think yourself, though. Sometimes the answer isn’t the most complex but the most basic. In this case, with four shots from the 3 yard line, that may have meant a dose of good old power football.

This is twice now the RPO bug has bitten us with the game on the line. Sometimes you have to learn a lesson the hard way before the correction is made. As far as the Steelers and RPOs go, let’s hope yesterday did the trick.