You wouldn’t think that’s a dilemna, but if you’re a political party, it is. Do you put it all in close races (the conventional wisdom), or spread it around a bit more? This article is the kind of analysis I love, using the power of mathematics, statistics, and clear thinking to reach a seemingly obvious conclusion. The gist is that you get diminishing returns from the big money thrown at close races. That money has more of an effect in races where your candidate is facing long odds. In any race, that seems foolish, but taken as a whole, some of those long odds will convert.
I would also note that part of the GOP strategy has been to attack all the Democratic strongholds. They’ve made huge inroads into Labor, the African-American block, the poor, Catholics (despite the Democrats running a Catholic candidate!), etc. Although these are all still majority blue, they are no longer overwhelming, and the Democrats have had to spend much time defending their home turf. Spending money in places where the GOP doesn’t expect to have to fight a battle can have a disproportionate outcome, as they have to spend resources defending rather than attacking.