I would imagine that the experts that come out with the preseason polls have done plenty of homework - returning talent, strength of schedule, how they fared last season (yeah yeah, new year, new team, but often times teams don't change THAT much, barring coaching changes or a much-ballyhooed QB transfer), etc. There isn't an exact science, but I'd wager that they're probably a lot closer than what we give them credit for.
What they can't account for are the injuries, suspensions, chemistry (or lack thereof), poor play, poor officiating, crappy weather, etc. that can ruin a game or even a season.
Well first, I'm not sure I can agree with even your initial point. The coaches certainly don't spend a ton of time evaluating all 100+ D1-A teams and slotting them into their top 25. Most of them don't even vote, they leave it to assistants, interns, etc. And those assistants and interns don't spend much time either, they have actual real jobs to worry about.
AP voters might be slightly better, but the obvious regional biases in their rankings are well documented over the years.
The computer rankings don't really "do homework" because they're largely straight algorithms.
I guess some of the other various publications might put more effort into their rankings, Phil Steele and the other ones out there. But those aren't the rankings that anyone's using when they advertise "Two Top 25 teams in the B1G square off this Saturday!!!!!" so really anything that's not AP, coaches' poll, or CFP rankings, is pretty irrelevant to the way matchups are understood throughout the season.
But even if we accept that these entities DO perform such due diligence, they're still only right about half the time, so it's a fairly pointless exercise regardless of how much effort goes into it.
But hey, we get to argue all preseason, and then all season long, about what some yahoo thinks of our teams, so I guess it serves its purpose anyway.
