Packers 14, Falcons 35 — Week 3: Nine Runs to Atlanta's Thirty-Nine

First and 10 at the Atlanta 40, three oh six left in the first quarter, Green Bay up seven to nothing. A roughing-the-passer penalty had just turned a third-and-11 incompletion into a first down for an Atlanta offense that had not run an offensive play inside an opponent's 20 in two games. Bijan Robinson went around left end for 55 yards to the Green Bay five. Nine points of win probability on one snap — win probability being the model's read on each team's chance of winning at that moment — and by expected points added, what a snap was worth given the down, the distance and the field position, measured against what that situation was already worth, +3.7. Four days earlier this run defense had taken 11.8 expected points away from 27 Jets rush plays and won a game with it. On Thursday it allowed 244 yards on 39. Green Bay lost 35 to 14 at Lambeau Field, led for less than four minutes of the first quarter, and trailed by at least 10 points for the entire second half.

Start with the run defense, because it is where last week and this week disagree. 39 Atlanta rush plays, 244 yards, three touchdowns, +12 expected points — on defense a big positive number is the bad direction. 64 percent of those runs were successful plays, meaning they gained what the situation asked for. Among the 66 team-games this season that is 60-third of 66 by expected points allowed per rush play and 60-fifth in rushing yards allowed; only Carolina's 291 against Chicago in Week 1 was worse. Green Bay's own worst was Baltimore in Week 17, 308 yards and +20; no other game this defense has played since the start of last season allowed more than a hundred 63. For the baseline: at the rate this defense allowed over Weeks 1 and 2, 39 carries would have been worth about −9. At last season's rate, when it ranked seventeenth, about −1. All three rushing touchdowns came from inside the 10.

The pass defense faced a quarterback making his first pass plays of the season. Michael Penix took the first snap for Atlanta after Cooper Rush and Jack Strand had handled the first two weeks, and 65 percent of his 26 pass plays were successful against this defense — the fourth-highest success rate any offense has posted in the 66 team-games this season. +9 and a half expected points allowed, 256 yards, one touchdown, one interception, no sacks. At last season's rate, when the pass defense ranked 20-third, 26 plays would have been worth about +2. What the unit got was the biggest swing of the night: second and 10 at the Green Bay 49, 13 33 left in the first quarter, nothing to nothing, Penix short middle for Jahan Dotson and Xavier McKinney at the 40 — −4.9 expected points, 11 points of win probability. What it allowed was Drake London, six times for 16 yards or more, the longest on third and two at the Atlanta 16 with 14 17 left, up 24 to seven: Penix deep left for 68, pushed out at the 16, +5.2 expected points, more than any other snap in the game.

The passing game opens on a count, because it is a record for this offense over two seasons: 54 pass plays, more than Green Bay ran in any game last year, and more than any team has run this season except New Orleans and Houston. They produced +3.6 expected points, 312 yards, two touchdowns, one interception, one sack taken, and 41 percent of them were successful plays. At the rate Green Bay threw last season, when this was a top-three passing game in the league, 54 plays would have been worth about +12. At the rate of Weeks 1 and 2, about −4. So Thursday was better than the first two weeks and under a third of last season. The best snap Green Bay had all night came early: third and 11 from the Green Bay 12, nine 18 left in the first quarter, Jordan Love deep right to Matthew Golden for 45 to the Atlanta 43, nine and a half points of win probability, +4 and a half expected points. That drive ended with Love to Christian Watson for four yards on first and goal, seven to nothing. The second touchdown came on second and seven from the 15, Love short right to Golden with eight 21 left, down 27 to seven.

The run game is nine rush plays for 17 yards and −2.7 expected points. Fewest rush plays and fewest rushing yards in any of the 66 team-games this season, and no team ran it fewer than 10 times in any game last season. The longest gained six; seven of the nine gained three or fewer. Per play it was better than either of the first two weeks, on a sample of nine.

Two snaps decided the shape of the second half, and both were fourth downs of a kind. A 34-yard pass interference penalty put Green Bay at the Atlanta 33 with four 44 left in the half, down 10 to seven. A five-yard run, two incompletions, and Trey Smack's 47-yard field goal was blocked by Zach Harrison with three 47 left — 10 points of win probability, −3 and a half expected points, the second-biggest swing of the game. Atlanta went 63 yards in eight plays for the touchdown that made it 17 to seven at the break. Then the opening drive of the third quarter: six plays from the Green Bay 30 to first and 10 at the Atlanta 16, then MarShawn Lloyd for two, Lloyd for two, Watson for five to the seven. Fourth and one, 10 oh two left, down 17 to seven, and Love's throw to Watson fell incomplete: 10 points of win probability, −3.7 expected points. Atlanta went 92 yards in 10 plays from there. Between the first touchdown and the second, six Green Bay possessions ended without a point: three punts, the blocked kick, the stop on downs, and another punt.

The drive chart, 10 drives: two touchdowns, five punts, a blocked field goal, a stop on fourth down, and the interception Xavier Watts took at the Atlanta three with one 18 left. Four three-and-outs, after five last week. Third down was six of 15, 40 percent — after three of 12 in Week 1 and one of nine in Week 2, the best third-down night of the season. Fourth down was two of three, and the two conversions came down 20 21.

For anyone new to this show, here is where that sits. Last season the Packers went nine, seven and one with a top-three passing game in the league by expected points per play, the second-best third-down conversion rate, a run game nineteenth, a run defense seventeenth and a pass defense 20-third, and the preview said the offense's case was settled and the defense's was the question. Week 1 was a 39 to 22 loss at Minnesota in which the passing game ranked nineteenth and the run game last. Week 2 was a 20 to 17 overtime win at the Jets on one third-down conversion in nine, carried by a run defense that took nearly 12 expected points off 27 carries. Thursday moved the offense a little and the defense a lot. The passing game had its best total of the three weeks and still finished under a third of last season's rate; third down went from 11 percent to 40; the run game was nine plays. And the run defense that won Week 2 allowed the second-most rushing yards it has allowed in any game since the start of last season. Three games do not settle a season. But the offense this team was built around has now been under last year's rate three times in three, and the unit the preview called the question came out of Thursday with a louder one.

Green Bay is at Tampa Bay on Sunday afternoon, October fourth, at Raymond James Stadium.

That's your Packers — every game, by the numbers. This is Muffed: we do the same deep dive for all 32 teams. Follow the Packers show wherever you're listening, so next week's recap lands the moment it's out. No theater. Just the football.

And if you play fantasy, there's a Muffed show built around your exact roster — free at muffed.ai.

_Data via nflverse._