As a longtime FPS player one area of complexity that is easy to forget is the maps.
When I play the game, I can learn a new map pretty quickly.
When I watch a game played on a map I'm not familiar with, it takes me much, much longer to develop the same understanding. It doesn't load into memory the same way. Similar to remembering the route to a new location as a passenger vs a driver.
The broadcast might flip back and forth between two players, hunting for each other in different parts of the map, and unless I know it pretty well I could be totally lost until they actually meet.
The broadcaster can help by using third person / freecam. But at the same time, an FPS loses a lot of the tension and visible skill when combat is viewed from freecam. So finding the right balance of camera time between individual players' views and an overhead view is challenging for a broadcaster, far more so than in traditional sports where we're accustomed to seeing a wide open field and perhaps some tracking of the ball.
Edit:
To give a practical example, I used to dabble in Quake Live. I was never any good, but there are some maps I know well: dm6, ztn, dm13, t7. Even though I'm not a big eSports fan, I don't watch streamers or whatever, I've seen some VODs played on those maps that really held my attention with suspense. I've been surprised at how entertaining it was watching top players duel.
At some point I watched a Quake Champions duel match between Rapha and Cooller and while the gameplay was quite familiar, not knowing the maps made it far less entertaining. I just couldn't follow the significance of the player's positioning nearly as well. Before that I totally underrated how much my familiarity with the maps was adding to my enjoyment of the Quake Live tournaments.