In 2018, the 25 busiest US airports served just under 1 Billion domestic passengers. At least that's the number reported by airports based on the most common calculations. When airports report their passenger totals, they count everyone that steps off of a plane into the airport or steps out of the airport onto a plane. So when a passenger flies between say Newark and St. Louis on a direct flight, they get counted 2 times as a passenger. Newark counts that passenger as a departing passenger when they leave the airport and St. Louis counts them once as an arriving passenger. Now if that passenger, instead of flying direct had a connection that went through Chicago, they would count as a passenger 4 times. Once at Newark, once at St. Louis, and twice at Chicago (once as an arriving passenger and once as a departing passenger).

Connecting traffic can add a big boost to an airport's passenger numbers. So when airports state they have 10M, 20M, or 50M, how much of those numbers are coming from people traveling to and from the airport's nearby city or surrounding area and how many are just passing through the terminal en route to some other destination? Charlotte Douglas International (CLT) has the highest percentage of connecting domestic passengers at a whooping 70.5% of passengers only using the airport to connect. Among airports with at least 5M annual passengers, South Florida International in Fort Myers (RSW) had the lowest percentage at a meager 3.0%. Atlanta's Jackson-Hartfield Airport has been the busiest airport for many years with 92.2M domestic passengers in 2018, but as the centerpiece of Delta's entire flight network 57.9M of those passengers were using the airport to connect to their final destination while only 34.3M were Origin & Destination (O&D) Passengers, meaning they began or ended their trip at ATL. While Atlanta and surrounding areas generate lots of passengers for the airport, the fact that Delta chooses to have their largest hub in Georgia is the main reason that Atlanta is always the busiest airport. If we removed all connecting passengers and only looked O&D Passengers for an airport, LAX (plus ORD, DEN, LAS, and MCO) passes ATL as serving the most domestic passengers.

Below is a chart showing an airport's total domestic passenger count on the x-axis and their percentage of passengers that are connecting traffic on the y-axis. The size of each bubble represents how much of their traffic is O&D traffic. Only airport with at least 5M total passengers in 2018 are shown.

The colored bubbles are a hub for one of the four major US airlines (American-blue, United-gold, Delta-red, or Southwest-green) or multiple airlines (purple fill with the biggest carriers color on the bubble boarder). For Southwest, I considered their operating bases as "hubs". From the graph, we can see that most of the busiest airports are hubs and that hubs account for most of the connecting traffic (as expected). But not all hubs are created equal. The x-axis shows traditional passenger count and any bubbles aligned vertical are considered a similar level of busyness by traditional passenger count even though the traffic those airport generate from the surrounding city can vary greatly. Let's look at the 6th through 11th busiest domestic airports which all fall roughly along the 40.0m passenger line on the chart above.

	Total Domestic	O&D	Connecting	Percent Connecting
Airport	Passengers	Passengers	Passengers	Passengers
LAS	44,432,222	36,027,440	8,404,782	18.9%
SEA	42,905,818	29,354,360	13,551,458	31.6%
SFO	42,195,162	31,574,770	10,620,392	25.2%
PHX	41,712,307	26,889,540	14,822,767	35.5%
CLT	41,592,965	12,270,770	29,322,195	70.5%
MCO	40,425,835	36,002,880	4,422,955	10.9%

While each of these airports served between 40 and 45 million domestic travelers in 2018, they had large differences in how many of those passengers were connecting traffic and how many O&D travelers were served. Orlando International (MCO) generated 36M O&D passengers while CLT generated only 12.2M. If American decided to drop CLT as a hub, the airport would likely see it's total passenger count plummet with its passenger totals looking closer to something like Nashville International (BNA) (12.4M O&D, 15.8M Total) or nearby Raleigh-Durham International (RDU) (10.6M O&D, 12.2M total). American had a hub in RDU which is a 2.5 hour drive east of Charlotte until the 90's and although Charlotte was already serving more total passengers than Raleigh in the 90's, it isn't too big a stretch to think Raleigh could be a major hub today serving 40M passenger with Charlotte flying 10-15M if only American had continued to build their hub in Raleigh while American's predecessor, US Airways, had wound down hub operations in Charlotte. While Charlotte's passenger totals are highly dependent on passengers connecting, airports like Orlando's generate most of their passenger.

Grouping the Airports

So let's break up the 65 airports that served at least 5M domestic passengers in 2018. I broke these airports into 5 groups based on their characteristics. Here are the groups:

• The Behemoth Hubs
• The Central Connectors
• The Coastal Draws
• The Balanced Hubs
• The Feeders

Anchorage (ANC) and Maui (OGG) aren't in the groups and could probably be pulled out into a 6th group of "Pacific Airports" with Honolulu (HNL) joining them, but I am holding them out for now

Behemoth Hubs

ATL, DEN, LAX, ORD

If you have flown more than a few times in your life, chances are you've been through one of these airports regardless of destination. They have the 4 highest domestic passenger totals and all have over 34M in O&D traffic. Each of these airports is a hub for multiple airlines and in 2018 they accounted for 30% of all domestic connections. If you took a connecting flight in 2018, there was a roughly 1 out of 3 chance it went through these airports! In general they generate a great amount of O&D travelers and couple it with high number of connections, these hubs aren't going away anytime soon and with infrastructure improvement on the horizon are poised to grow even bigger.

Central Connectors

CLT, DAL, DFW, DTW, IAD, IAH, HOU, MDW, MIA, MSP, SLC, STL

This group is made up of airports mostly located in the center of the country (MIA being a notable exception) that have high connecting percentages considering their O&D traffic. All generate more than 30% of their passengers from connecting traffic. These are domestic workhorse airports which feature several "smaller" domestic hubs (DTW, MSP, MIA, SLC), a few airports that split their market with another airport (HOU, IAH, DAL, DFW, MDW, IAD) and STL which has the highest connecting percentage by far of any non-hub. Looking at the airports that split their market within this group, if DFW and DAL were together they would move into the Behemoth Hub category and if IAD combined with DCA, they'd be a Balanced Hub. This group of airports is not as invulnerable as the Behemoth Hubs to changing trends in the future. As smaller planes become more efficient and extend their ranges, will more direct flight bypass these central hubs or will they convince airlines to drive even more capacity through these airports?

Coastal Draws

BOS, DCA, FLL, LGA, MCO, PDX, SAN, TPA

Each of these airports sits on or near a coast (LAS is a bit of a stretch, but we are counting it) which reduces their effectiveness for domestic connections with all below 19% of domestic traffic coming from connections. However, each of them draw solid O&D number (all over 15M) driven by warm-weather vacation spots (MCO, FLL, TPA, LAS, SAN), the Northeast Megalopolis (BOS, LGA, DCA), or the allure of the Pacific Northwest (PDX). These airports do not make sense geographically for lot of domestic connections, but often have sizable international connections with the exception of LGA and DCA which have limitations on their allowed international flights. All of these airports are major draws for O&D passengers even though several aren't hubs for any major airlines. This group's growth is heavily dependent on their surrounding metros continuing the grow as destinations since they are unlikely to add lots of domestic connections due to their geography.

Balanced Hubs

BNA, BWI, EWR, HNL, JFK, PHL, PHX, SEA, SFO

These airports all fall between 20 and 36% connecting traffic and generate solid O&D numbers. They each balance their domestic traffic and connections with strong international connections, not relying too heavily in any one area. While they offer good options to connecting passengers, their surrounding metros support them as good passenger drivers. With growth from their metros or a push from airlines for more connections these airports seem poised for growth, but several locations could be limited from a domestic connections perspective due to geography.

Feeders

The Rest

These are the 30 remaining airports who's O&D traffic fall 4 and 13.5M with none having connection percentages over 15%. The only hub in this group is Southwest's smallest domestic operating base at Oakland. The rest receive a smattering of connecting passengers, but are generally lots of flights away from reaching a hub status. They offer many flights to the other hubs shown here which locals use frequently as connecting passengers effectively feeding the hubs. In order to have a chance of becoming hubs, most of these airports will need to first grow their O&D traffic enough that the amount of flights start to allow more connections to make sense. The two airports that appear closest to moving out of Feeder status are Austin (AUS) which has the highest O&D of the group at 13.4M and has been growing rapidly (and is centrally located in Texas) or Kansas City (MCI) which has the highest connection percentage of the group at 14.7% (along with 10.1M O&D). MCI terminal layout has hampered it from a connection perspective, but a new terminal in the works may alleviate some of those concerns in the future.

---

Hubs get placed at airports for a multitude of reasons, but an important factor in adding flights (connecting or not) is O&D traffic generated by the area surrounding an airport. Airport's with high O&D numbers are more likely to get even more passenger traffic as airlines funnel connecting flights through the airport. I have changed the plot we have above to show O&D traffic on the x-axis (and bubble size showing total traffic now), and plotted a line showing the linear trend of O&D Passenger Counts to Connecting Percentage of Total Passengers. The trend shows that the percentage of connecting traffic increases as O&D traffic increases. For an example, say you have 10M O&D passengers and have 18% connecting traffic (yielding 12.2M total passengers). If you added 1M O&D passengers and your connecting traffic percentage stayed the same you'd have 13.4M total passengers, but the trend says that all other things being equal, the airport would likely see its connect percentage go up about half a percent to 18.5% meaning that the airport would have 13.5M total passengers getting another 100k passengers due to the higher likelihood of receiving additional flights catering to connections.

A couple of significant factors that are not accounted for in the trend are that 1) we are looking at only domestic traffic and not any international traffic 2) since we aren't considering international passengers many of the coastal hubs see lower connecting percentages than at more central airports because the international connections aren't being counted. The data used for all these analyses was obtained from T-100 Segment (total passenger counts) and DB1B Market (O&D passenger counts) data and looks only at domestic travelers. When spot checking the passenger totals against domestic totals reported by airport while the number aligned the T-100 data was generally slightly lower than the airport reported numbers. Please dig into the interactive graphs above!

---