Co-founder and long-time chief executive of Southwest Airlines
Made an airline profitable for decades by refusing to add things customers were not paying for.
Photograph: SouthwestArchive · CC BY-SA 4.0 · Wikimedia Commons
Southwest began flying in 1971 after years of legal fighting from competitors trying to keep it grounded. Kelleher, a lawyer by training, ran the defence and later ran the airline.
The model was ruthlessly simple and every part of it reinforced the rest: one aircraft type, so training, spares and maintenance were simpler; point-to-point rather than hub-and-spoke; no assigned seats; short turnarounds, so an expensive asset spent more of its day earning.
Any single element could be copied. What was hard to copy was all of them together, plus a workforce culture that made fast turnarounds normal rather than heroic. Full-service airlines that launched low-cost subsidiaries generally struggled, because they kept enough of the old model to lose the advantage.
Southwest's advantage came as much from the meals, seat assignments and interline connections it refused to offer as from anything it added. Most small businesses are the opposite: they say yes to every request and end up with an operation too varied to be efficient at anything. Write down what you do not do.
One aircraft type made everything downstream cheaper. In a smaller business the expensive thing might be your vans, your machines or your software. Every extra variant multiplies training, spares, error and cost — and the saving from buying a cheap odd-one-out is usually smaller than the ongoing complexity it creates.
Turning aircraft around faster meant more flights from the same asset — more revenue without more capital. Before discounting to win volume, ask how much idle time your expensive assets and people have, because filling that is margin you already own.
The model constrains what the company can become. Southwest spent years working out how to handle assigned seating, long-haul and international flying without unpicking the economics that made it work — and later had well-publicised operational failures when its systems were tested hard.
A tightly-coupled model is efficient precisely because it has little slack, and little slack is exactly what you need when something goes wrong. That trade is worth making consciously.