College football's average TV audience is down 11% this season, by Nielsen's count. Fox Sports says total viewing is up 2%, because more games are on national television. Nielsen's own count says there are eight fewer games. Both numbers are being quoted as the state of the sport, and on the bases they were published on, they cannot both be right.

Start with Nielsen. Through Week 9, FBS game broadcasts are averaging 1.6 million viewers across every Nielsen-rated network, down 11% from the same nine weeks of 2025, according to Nielsen data provided to Front Office Sports. The comparison covers 402 games this season against 410 last season. The September number was the same: down 11%, at 1.7 million.

Now Fox. Michael Mulvihill, Fox Sports' insights and analytics chief, says total minutes viewed across all networks are up 2% through nine weeks. He laid out the logic on X at the end of September, when total time spent on college football was up 0.2%: the average is down, he wrote, but the number of games on national networks is up, so "the higher volume negates the lower average viewership." To him, total viewing time is the most correct way to read a sport's trend.

That is a fine principle. The arithmetic doesn't support it here.

Total viewing is the average audience times the number of games times how long each game runs. Take Nielsen's published figures at face value: the average is down 11% and the game count is down from 410 to 402. Multiply them and total game viewing should be down roughly 12.7%, not up 2%. For the two numbers to reconcile on the same set of games, the audience-weighted length of a broadcast would have to grow by about 17% in a single season. Nobody has said college football games got 17% longer.

So the 2% is counting something the 11% isn't. Different networks, different games, streaming-only windows, studio shows, or a different way of averaging. Fox hasn't published which. Until it does, "up 2%" is a number without a denominator.

Frank Kowalczyk: Negates. The volume negates the average. Negate means cancel. It doesn't mean add. Down eleven, minus a couple of games, is not up two. I did this math on scanner data in Northbrook in 1986 with a pencil and a guy named Sal. Sal would have caught it before lunch.

The 11% decline is measured with a more generous instrument than the one used a year ago.

Nielsen has made three major changes to how it counts audiences in less than two years: expanded out-of-home measurement in February 2025, Big Data + Panel in September 2025, and, effective Aug. 31, 2026, a package led by a new co-viewing methodology that captures more people watching together. Sports saw 10% to 15% increases and a run of records in the 11 months after Big Data + Panel arrived, Front Office Sports reported. The co-viewing pilot in February produced an average 4.19% lift across the Super Bowl, the Olympics and other marquee events.

The Athletic's Stewart Mandel, analyzing games on ABC, CBS, NBC, Fox and ESPN through Week 8, found Nielsen's latest changes have bumped baseline viewership by about 3.4%. Those five networks are averaging 3.3 million viewers, down 6% from 3.5 million.

Apply that 3.4% to Nielsen's all-network figure and, by Unmeasured's rough arithmetic, college football on last year's ruler would be down closer to 14%. Not a crash. Not noise either. The methodology is holding the number up.

Frank Kowalczyk: Three new rulers in two years. Every one of them made football longer. Not one made anything shorter. Forty years in measurement, I've never seen a ruler do that. Fishermen do that.

Fox's math can work. It just has to be counting a different slice of the sport than Nielsen. There's a good candidate, and it tells you who the 2% is for.

Below the line: the subset that makes Mulvihill right, the one network that grew, the tier of games the 24-team playoff is supposed to sell and how far it fell, and why the NFL called Nielsen's new ruler a negative for sports right before it added 3%.

Enter your email to keep reading. It's free.