We researched every completed Lions season from 2001–2025, then compared the Millen, Mayhew, Quinn and Holmes eras across the same seven football questions used in Report 001. The raw workbook stays private; the rules, evidence inventory, source policy and weaknesses of the model are public here.
Start here
Millen is scored from 2001–2008, Mayhew from 2009–2015, Quinn from 2016–2020 and Holmes from 2021–2025. The 2008 transition is treated as a Millen-built season because he constructed the opening roster before being fired after Week 3.
Seasons, playoffs, quarterbacks, draft classes, notable picks, major roster moves, coaching decisions, elite player seasons and major adversity cases all get their own structured evidence rows.
Winning matters most, but front offices also have to solve quarterback, find and develop players, allocate resources, align coaching and survive when the plan breaks.
A GM does not get the same credit for creating a solution, inheriting it and maintaining it, or converting it into the next solution. Detroit made that distinction especially important at quarterback.
We change the weights, remove one category at a time and run random-weight models. If the winner depends on one preferred formula, the public verdict has to say so.
Seven football questions
The published weights are the same NFL framework used for Indianapolis so the reports remain comparable.
Regular-season win rate and the rate of producing winning seasons.
Playoff appearances, division titles, deep runs, Super Bowl trips and championships per season.
Objective starter stability plus an explicit interpretation of who created, inherited or converted the quarterback solution.
65% mature draft index and 35% elite player/homegrown-core index. Detroit's 2023–2025 classes are tracked but deliberately excluded from mature class grading.
Major trades, signings, extensions and exits, using average move quality, positive-move rate and catastrophic-failure avoidance.
Coaching continuity plus major hiring, retention, firing and staff-development decisions. Ownership-led choices are not falsely credited to the GM alone.
Preparation, replacement execution, result preservation and in-season adjustment during major availability shocks. We grade the response, never blame the organization for an injury.
Mayhew used the No. 1 pick to create a 12-year franchise-quarterback answer.
Quinn gets stability and retention credit for Stafford, but not acquisition credit.
Holmes gets succession credit for turning Stafford's exit into another long-term quarterback plus premium assets.
Evidence integrity
328 of 328 core records have row-level source coverage. Primary source families include Detroit Lions official material and Pro Football Reference, with other reliable sources used for event-specific cross-checks when needed. Source coverage makes the evidence auditable; it does not make judgment-coded grades indisputable.
Can the formula be blamed?
Why? Holmes currently leads all seven underlying dimensions. With nonnegative weights on those same seven categories, another era cannot pass him unless the evidence or category scores themselves change. That is a robustness result about this model—not a claim that Holmes has a 100% probability of being objectively correct.
What the model cannot claim
The Holmes era is the strongest in this 2001–2025 comparison without yet reaching the Super Bowl.
2023, 2024 and 2025 are visible evidence but excluded from mature class averages. Early success is not turned into a permanent grade.
The 9–8 regression and another injury-heavy season prevent the turnaround from being presented as invulnerable.
Move grades, adversity-response scores and some category construction require documented human judgment. Sensitivity testing shows whether reasonable changes overturn the conclusion; it does not eliminate subjectivity.
Why isn't the entire workbook public? The native 28-tab research database is private so the organized work product is not simply handed away. The public site exposes the evidence inventory, formulas, cutoffs, source policy, limitations and robustness results needed to understand and challenge the conclusion.