I suspect they are probably doing some validation after they push their changes. I would hope they are. I doubt they are running on Docker containers and probably still on a Windows server infrastructure. We also aren't talking about the Google, Meta and Amazon of the IT world.
I run an application on a Windows Server OS that takes about ~ 10 seconds of downtime to update itself. Granted, this is a commercial product, but still. Even our in house webapps, they do run on Alma 9, not containers, but they're again a couple second push to update the code. Unless you're changing out the underlying OS, there's no reason to need these sorts of outages to update things like a expansion menu not working. We can push changes 5 times a day and users don't notice cause the outage is like a second.
And even if you're changing out major stuff on the back end, it's not exactly rocket science to do so with much shorter outage windows - you stand up the new set of OSs, VMs, DB servers and once you've validated them you change the HA method to point to to the new rather than the old ones. Again, split second updates to the end user.
I know I don't know what I don't know but this is pretty insane for them to regularly have day long IT outages for their website. I work for what's basically a SMB, it's not like you need a FAANG level infrastructure team to do this. This all smacks of a part time IT guy basically lol. We have maybe 4 FTEs for the services I'm talking about?