An example of this is RDU, where AA and DL now have a fairly similar market share at 26% and 30% respectively. However, Delta offers 27 non-stop destinations compared to AA's 12. At some point the frequent flyers out of RDU will start to question why they fly with American, when Delta is much more likely to overfly their hubs and provide their RDU base with shorter and less complicated commute.
The tradeoff is that if a carrier has a disproportionately more "convenient" (more frequency, more seats, etc.) offering in the markets that represent a disproportionate share of that O&D demand, the incremental marginal revenue generated from catering to the "long tail" of smaller O&D markets may not be worth the incremental marginal cost.
That is, in a nutshell, the divergence in domestic network strategy between AA/United and Delta. As said previously, both AA and United have hubs in many of the metro areas that often represent the largest and most important O&D markets for just about any non-hub spoke in the country. The same cannot be said, or at least cannot be said to the same extent, for Delta. So in that context, it isn't hard to see why both AA and United have concluded that the profit-maximizing approach is to leverage the economies of scale and market penetration they have in these massive O&D markets to pull local demand in non-hub spokes onto their respective networks, and accept that this may mean losing some business in non-hub spoke markets from the subset of customers who absolutely demand nonstop access on thinner routes.
Put differently - to the example above: AA seems to have concluded that there is enough high-value traffic to be captured in RDU from corporate customers who fly constantly to NYC, WAS, CHI, etc. and are willing to connect for smaller markets, and can be served on AA's multiple daily nonstop flights on all those routes that have relatively lower marginal per-seat cost because "the plane was already going there anyway," and thus if AA loses a few RDU customers who must
have nonstops to CMH and BDL, then so be it, because the marginal cost of flying a plane to those cities just to cater to those few customers isn't worth the revenue potential.
Running a city pairs through a hub may increase revenue, because of the other city pairs you can sell, but it increases costs (which affects profits) and reduces demand from increased travel time.
It's not that simple. The key isn't whether one alternative "increases costs" - in the abstract - versus another. The key is "compared to what
," or opportunity cost. All else equal, sure, if there is already a plane overflying the hub between two spokes, the incremental marginal cost of a seat on said plane is likely less - at least in cost-based, if not always value-based, terms - than two seats connecting over the hub. But there are undoubtedly innumerable examples of flights operating through hubs - especially hubs like AA and United have that tend to be in massive O&D markets - where the incremental, marginal cost of selling seats on two planes in a connecting city pair is almost certainly lower
than selling a seat on a nonstop flight in that city pair. Economics of scale are real, and they're exceptionally powerful.