The tidy story about Kevin O'Connell and J.J. McCarthy was supposed to write itself. O'Connell inherited a 33-year-old Kirk Cousins in 2022 and got one of the best passing seasons of his career out of him. He turned journeyman Nick Mullens into a plus passer for a month. He handed Sam Darnold a career year in 2024. And then, finally, the "quarterback whisperer" was going to get to work on his own guy — the QB Minnesota traded up for at pick 10 in 2024, redshirted for a season, and installed as the starter in 2025.

Instead, 2025 was the year every version of that story fell apart.

The system worked for everyone else

Here is every quarterback who has thrown at least 30 dropbacks for Minnesota under O'Connell (regular season plus playoffs), ordered by season:

Every quarterback O'Connell has trusted with real volume — Cousins twice, Mullens in relief, Darnold on his one-year audition, and Wentz picked off the scrap heap in September — has thrown for a positive CPOE. Only Josh Dobbs, who joined mid-October 2023 five days after being traded and started twice while learning the playbook on the fly, was negative. And the incumbent the offense was reportedly built around posted a −5.72 CPOE and −0.198 EPA per dropback on 273 attempts, with 11 touchdowns, 12 interceptions and 27 sacks.

Among the 38 quarterbacks who threw at least 200 dropbacks league-wide in 2025, McCarthy's −5.72 CPOE was the second-worst in the league, ahead of only Cleveland's Dillon Gabriel (−7.51). By EPA per play, McCarthy's −0.198 ranked third-worst in that group — behind Cleveland's Shedeur Sanders (−0.299) and Gabriel (−0.220) — but it was still worse than Tennessee rookie Cam Ward's −0.192. On the accuracy measure that matters most for a passing offense, McCarthy was outperformed by all but one qualified quarterback in football.

The Wentz control

The reason this piece exists — the reason "McCarthy struggled as a rookie" isn't the whole story — is what Carson Wentz did on the same field with the same weapons for a chunk of the season.

Wentz arrived in Minnesota after four straight negative-CPOE seasons the last time he'd been a real starter: −0.77 in Philadelphia in 2019, −3.49 in Philadelphia in 2020, −2.12 in Indianapolis in 2021, and −3.75 in Washington in 2022. Between those seasons and 2025 he'd thrown exactly 50 total pass attempts, as a fourth-stringer in Los Angeles (2023) and Kansas City (2024). He turned 33 during the 2025 season. Nobody signed Carson Wentz expecting a plus passer.

On 188 dropbacks in Minnesota's offense, he was one — +1.11 CPOE, +0.034 EPA/play. That's his best CPOE in any season with real playing time since his 2018 Pro Bowl year (setting aside those 50 cameo attempts in 2023-24). Not All-Pro numbers. But a career-negative-CPOE journeyman posting his best accuracy mark in seven years in the exact scheme where the first-round incumbent posted the second-worst CPOE in football is the sort of coincidence that isn't one.

The controlled comparison: same target, different QB

The cleanest test in football is when two quarterbacks throw to the same receiver in the same offense in the same season. Minnesota gave us that test twice over.

Justin Jefferson, targeted by:

Jordan Addison, targeted by:

The gap on Jefferson targets — CPOE +3.81 vs −9.59 — is 13.4 points. The gap on Addison targets is 22.4 points. And McCarthy wasn't attempting a wildly deeper passing tree on Jefferson: his average air yards to Jefferson (10.2) were actually shorter than Wentz's (11.0). Only on Addison did McCarthy push it noticeably deeper (15.2 vs. 12.5 air yards), and that's exactly the receiver where the CPOE gap was widest.

Same coach, same play-caller, same receivers on largely the same route tree. One QB threw them plus balls; the other threw a double-digit-negative CPOE at the best receiver in football.

What the workload split suggests

McCarthy carried the bulk of the season's dropbacks (273) as the season-long starter bookending everything else; Wentz's 188 came in relief over a shorter midseason stretch; Brosmer mopped up the final 85 as the emergency third option. That is not a small-sample fluke buried inside a good season — it's the offense's primary quarterback posting the league's second-worst CPOE across the largest passing workload on the roster, while the fill-in outproduced him on efficiency with fewer chances. Whatever the team's full-season output looked like in the aggregate, it was built on top of that split.

What I'd read into it

To be fair to the sample: 273 dropbacks is a season, not a career, and CPOE noise on that volume is real. McCarthy is 22 years old, made one full season of college starts, and is effectively a year removed from a 2024 meniscus tear that cost him his rookie season. This isn't a verdict.

But the "O'Connell will unlock him" argument leaned specifically on the idea that this offense makes quarterbacks better than they'd otherwise be. What 2025 actually showed is that the offense turned Carson Wentz — a QB who hadn't posted a positive CPOE in a full season since 2018 — into a plus passer for two months, while the quarterback it was built around finished with the league's second-worst CPOE among qualified passers.