Can someone enlighten me on what logic is going on behind the scenes?

If you search SFO-DEN for 12/24:
- 5:15am departure has H9, Q9, V9
- 6:00am departure has H9, Q9, V5
- 2:30pm departure has H9, Q9, V5.

The only DEN-YYC flight on 12/24 is at 7:03pm and has Q9, V6, W2.

As expected, searching SFO-YYC has V9 for the 5:15am option (10+ hr layover!), V5 for the 6:00am option (9.5 hour layover). However, for the 2:30pm option, it shows U9, H1, and prices at close to $567 rather than $368 despite both component segments having much better availability. Is UA purposely using married segment logic to make shorter connections book into a higher fare bucket and more expensive?

Multi city search priced at $380.

No, you mistake their intentions.

They are purposely using married segment logic to right-size the through inventory on SFO to YYC to match sales and demand on that citypair.

The only nonstop on that route is currently J2 Y3, and one of the two DEN-YYC flights (11:24a departure) is already sold out. The fact that you can book earlier flights without the married inventory suppression and without a fare break is due to an oversight in your favor in the fare rules, which specify:



Because the earlier flight is sold out (or sold out only in that inventory class?? seems to work), it is permitting the longer connection without a fare break even through the inventory algorithm "forgets" to marry the inventory.

Ah, thanks for the info. So by default, does the inventory algorithm marry things to the next departing flight? So in this case, are the 5:15am and 6am "married" to the 11:24am departure, while later SFO-DEN flights are married to the 7:03pm departure, which is why the earlier flight being sold out is allowing the 5:15am and 6am ones through for the later flight.

I believe the issue is that the inventory algorithm only thinks the segments are married if the connection is less than 4 hours, because that is supposed to be the rule for transfer-versus-stopover on a transborder flight. If that rule held, the longer connections would not be married but there would be a fare break in DEN, pricing as SFO-DEN plus DEN-YYC the same as if you purchased two separate tickets.

I actually don't know why it is pricing the longer connection with a single fare, because it works on days where the morning flight is not sold out, and even in fare classes where the morning flight has available. I don't know if the "12 hours" rule in the fare template text is being misread by the computer, or if there was a pandemic change to the transfer time for transborder flights. Perhaps jsloan can help figure it out

My broader point would be that, on a day to day basis, nobody at UA knows or pays attention at this level of detail. Someone went into the computer and coded that SFO-YYC via DEN will have suppressed inventory for whatever business reason, and didn't notice or consider the weird long-connection thing.

My thought as an amateur here are that UA is charging more for the desirable itins because more people have booked them. I don't think it's anything nefarious.

Not that you've asked but AA has done this for many years on their awards. For example, I can get a 20k first class trip to Cabo if I double connect and overnight in PHX. For 6 hours! If I want a sensible itin, it'll cost my 90k. UA is not quite so literal with the cash to miles ratio but they are definitely trending toward 1 cpm unfortunately! I still don't quite understand how the pricing engine determines award fares and my friend has not been forthcoming. Hoping to jump him at the next meetup if we ever get one!!!

It might not just be for award tickets. I know that this thread is about UA, but it might also explain what I saw recently when I purchased a SJC-LAX-MIA-SJU round-trip, which AA was offering for a remarkable $123 round-trip.

The outbound SJC-LAX flight had a nine-hour layover before the LAX-MIA red-eye, in spite of their being a later SJC-LAX flight that would require a much shorter three-hour layover. I assumed that the cheapo seats were gone on that later SJC-LAX. But then I saw a similarly priced routing with the later SJC-LAX flight -- but it required an overnight at LAX before continuing on to MIA the next morning. When I forced the computer to combine that later SJC-LAX with the LAX-MIA red-eye, the price went way up!

To the layman (me), it appears that AA is publishing a very low price but charging much extra for the most-desirable flights with the shortest connections. There might be some truth to that, but it could be as findark explained, and the computer didn't apply the married segment logic to the flights with a longer connection but also didn't break the fare.

Yeah, I think many of you are reading malicious intent into random noise.

Specifically with regard to the price of A-X-Y-Z-B with a redeye and an overnight when compared to A-B nonstop... usually the convoluted route is more expensive (for obvious reasons), and sometimes it ends up being cheaper (glitch, nonstop is really expensive, those flights are each really empty, etc). Especially because for given A-B there are going to be many such complex itineraries and your search is only going to unearth the one that is cheap. There is no reason and no benefit for AA to sell e.g. SJC-LAX-MIA-SJU with worse connections for cheaper than with better connections, unless the individual flights just happen to emptier, at less desirable times, etc. as compared to a "better" routing. And there was probably a poorly timed SJC-ORD-DFW-SJU you never considered because it was more expensive.

AA absolutely does this. The longer I choose to sit around DFW, the cheaper the fare, regardless of when I leave (early or later) or the flights I'm on. If it's a double connection, it gets even cheaper - then you just have to decide which of the two airports you want to sit around for 4 or 5 hours. I have no idea why they do this.

What UA is doing is a bit different where they are offering combinations that normally wouldn't come up as a sensible option. I saw this the other day with fares laying over in Denver that were just end on end combinations of very cheap CLE-DEN and then DEN-PHX (for example) flights, but the end on end rules seemed to prohibit a connection time of less than 4 hours. (You could purchase the two trips on two different tickets/records and have a shorter connection.)

I guess the "next available" wording is a red herring, and the provision should be read like you would read "if passenger takes next flight within 12 hours," with "next flight" not "the next scheduled flight" but rather "the next flight that the passenger takes." "Available" then prevents using an unavailable flight to avoid constituting a stopover. US domestic fares carry identical language but usually say 4 hours.

But this is about how autopricing parses this provision, and I'm puzzled. How did you get this display? EF shows that AVAILABLE is deliberately indented on the next line, which is noteworthy as there is still sufficient room to keep AVAILABLE on the previous line, and nowhere else are running autopricing provisions indented when they reach the next line.



If it is actually there, I believe the extra indentation splits "next" and "available" in two different conditions for autopricing to evaluate. "No stopover occurs" passes if both are true, meaning [(an) [i]available flight within 12 hours] [is the passenger's [i]next segment], which matches what we see in the field and is in line with my take on this provision explained at the beginning.

I suspect the formatting thing is a red herring. This is how ITA Matrix formats it. I think your interpretation is incorrect if you are saying what I think you are saying. As an example, UA only has 2x daily flights on SFO-ASE and fares to ASE only allow 4 hour connections before stopover is created. There are numerous cases where next available flight is over 4 hours when connecting through SFO to ASE (BNA-SFO-ASE, for example), but UA won't price these on a single fare. EF shows married inventory on these long connections, but you can't actually purchase them on a single fare. I suspect people are reading a little too much into married inventory. I've searched for inventory on DTW-MSP on DL a number of times and EF will often show married inventory for DTW-CVG-MSP connections. However, when I look at DL fare filings on EF, all DTW-MSP fares require non-stop routing and you can't actually purchase the CVG routing on a single fare.

That's routing though; BNA-ASE is via BNE-HOU/DEN/CHI-ASE. I'm fairly sure inventory will pull the through-inventory rule if the connection is short enough to ordinarily count as a transfer (4h00 domestic or next scheduled flight) or not otherwise (regardless of fare rules). You might be right that the fare rules on SFO-YYC actually auto-price like "up to 12 hours, no stopover", just a weird wording. That does seem to be how the computer does it. My original point was that there is no inventory marrying rule for the >4h00 connections on SFO-DEN-YYC because that part of the computer thinks it's impossible to price on a single fare. The fact that assumption is wrong is the whole reason why the situation in the OP came up.

I agree with your observation that inventory validation seems to be driven by its own rules and is oblivious to what the fare rules are saying about connections. I don't know if this is deliberate or just an oversight of the "this is never supposed to happen"-kind, but I agree it's more likely that no one noticed or considered this. Nice find and a good thing for OP!

I know this isn't new, but recently I've been seeing it impacting almost every connecting flight that I've booked.

UA seems to be frequently limiting availability on married segments as a way of charging more for shorter connections, relatively to longer ones.

For example, I'm looking at TLV-EWR-SFO in a few weeks. The obvious connection is the one with 2h10 between flights, however as a married segment this pair of flights is S9T7L1K0, so a minimum L fare for one passenger ($70 more than K), or S/T for two (doesn't appear to be a T fare filed). The 5h40 connection option is T9L4K1, and you need to go even longer before you get to K2 or higher. ie, the shorter, more convenient connection is more expensive than the longer connections, made worse if you're not travelling alone.

The underlying flights are completely different - S9T9L4K3 for the TLV-EWR, and S9T9L9K9 for all of the EWR-SFO flights. Splitting the married segments (ie, multi-city) happily books the lower through-fare K price on any of the connecting options. Even tools like ITA Software Matrix don't seem to find these potential splits.

That's because it's not supposed to work, and UA can fix it at any time.