DFW and ORD: American’s Two Problem Airports

Air travel is like all other consumer services, providers try to 'sell' everything around the product, which influences consumers into thinking they're buying something special. Air travel consumption boils down to two fundamental items: cost and time, for leisure travelers the focus is cost and business travelers focus on time. The proof of airlines really selling these two is that the airline industry sells schedule and prices around that. To try to understand how an airline's schedule is working (that's the #1 internal issue) we modeled the US airline industry using the US DOT's On-time data along with several custom KPIs. The result is a fascinating peek into the business. Our first analysis is on American Airlines. The table below is a key item to bear in mind as we delve into the various data items in this post. That cost per minute explodes when multiplied across the 'lost minutes'. We have been tracking this metric for a long time. [caption id="attachment_197124" align="aligncenter" width="523"] US DOT; AirInsight[/caption] The Data We are starting with a big display, and clicking on it makes it larger. You are going to want to do this. [caption id="attachment_197126" align="aligncenter" width="640"] US DOT; AirInsight[/caption] The data comes from DOT's On-Time dataset. We are reporting data through June, so a half year. The focus is on the airline's top ten origin airports. We've used color to help make it easier to grasp what we are showing. Metrics The dashboard has a left/right split. On the left we have high level information that identifies the airline's top ten origin airports in the first chart. The second chart shows how each of these airports are functioning over the year. In the case of American, its DFW concentration is rising. The right side of the dashboard breaks down the various metrics and KPIs by airport. Green is better than red. American's top ten airports account for 61% of its domestic flights. This is concentrated, but not unusual compared to its peer group. These top ten airports have seen over 6 million 'recovered minutes' through June - Block Variance Minutes measures the difference between scheduled gate-to-gate time and actual operational time, calculated as actual elapsed time minus scheduled elapsed time. Negative values show that a flight completed faster than its scheduled block time, often due to favorable tailwinds or schedule padding. Because airlines pad their schedules to look good, a better metric is Schedule Recovery. Positive values indicate that the flight arrived closer to schedule than it departed, exposing how much padding or buffer the flight utilized to recover time. At the bottom left of the dashboard note the difference between Average Block Variance and Average Schedule Recovery. That's quite a gap and illustrates what airlines sell and actually deliver. You thought they recovered an average of 13 minutes, but really it's closer to one minute. The first table with costs per minute now comes into focus. Multiplying the cost/minute against lost time blooms into a huge number. Looking at the upper right table: Operations are highly concentrated; the top three airports account for close to 60% of operations. Median Station Delay - DFW is a problem because that number is minutes and at $106/minute there are hard costs. Notice ORD (Chicago) is also in the red by a large number. If you were wondering where American is hurting, you now know their two primary spots. Station Disruption Rate - DFW and ORD again, and you would not be surprised. Station Performance Score - DFW and ORD show zero improvement over last year. The bright spots are CLT (Charlotte), PHX (Phoenix) and PHL (Philadelphia). Now for the lower right table: Average Schedule Padding - LGA (La Guardia) jumps out as a problem. Most other airports are better padded. Average Propagated Delay measures the average amount of time a flight is delayed due to upstream network disruptions—specifically, the cascading delay inherited when an assigned aircraft tail number arrives late from a previous flight. High values signal a brittle flight network where morning bottlenecks at primary hubs cascade and multiply through downstream station rotations, while low values indicate strong network recovery and effective tail-swap buffering. Airports with better weather shine. Average Schedule Recovery Minutes are detailed here compared to the system-wide 1.1 minutes. Bottom Line American Airlines is struggling at its two key airports. DFW and ORD account for the bulk of the network's real disruption cost, and neither shows improvement year-over-year — that's not a rounding error at $106/minute, multiplied across millions of minutes. But its other hubs, especially CLT, show good results. PHX and PHL are doing relatively well. The bigger story here isn't any single airport — it's the gap between what American's schedule claims to deliver and what it actually delivers. A 13-minute average block variance sounds like a airline that's recovering time lost to delays. A 1-minute average schedule recovery says otherwise. That gap is mostly padding, not performance, and it's the same dynamic playing out station by station: the airports that look good on paper aren't always the ones running well operationally. The next question is whether American's investment in padding at DFW specifically pays off, or whether it just moves the problem from the on-time statistics to the cost statistics. We'll be watching the second half of the year to see if CLT's model — genuine recovery, not just buffer — can be replicated at the two hubs that need it most.

No results