A Battery of Confusion

Like a year ago I took home an APC SUA750XL that was no longer being used. Fun fact, it was running not one but two laser jets...which it says not to do multiple times on the unit! So there is that.
It sat in storage for so long because they are SUPER proud of the OEM batteries. $200+ for an RBC7 18Ah 24v battery. Aftermarket, which I have had good experiences with in the past, are substantially cheaper. But give the unit had a pretty rough time in it's previous location I figured it prudent to go with an official battery for it.
Battery was delivered last Saturday (8.30.25) and I promptly installed it in the unit, no load, and allowed to charge while I was off at a birthday party. 6~ hours later I got home and moved the unit in place of another, a APC BX850M, that has been running my network stack and main server.
This is probably my biggest crime here, I guess.
It initially was flashing the low battery signal, which had me pretty confused. It eventually stopped doing that and based on the results in the terminal I thought all was well.
Until yesterday.
I had to unplug the unit for a brief period of time, 20-30 seconds. In that small window it supposedly went from 100% battery to 7%! Blasphemy!
So I'm lead to believe it's a calibration issue with the internal monitoring. Funnn. I didn't really want to run a calibration tbh. According to APC they want roughly a 30% load on the unit during this, which is around 160w roughly, and that just so happens to be the load my setup puts on the UPS while idle.
It couldn't possibly complete a calibration with that amount of load on it, I reckon, given the run time was pitiful. 4-5 minutes tops. I have my main server set to gracefully shutdown after X minutes of power loss and it was too far out to even run.
So I swapped the two APC units again, since I know the 850M can at least give me 30~ minutes of network run time in the case of an outage (which, in our area, is becoming increasingly common). The hope, given that the cell in the 750XL is clearly larger, was to increase that run time (more on this in a bit).
After swapping them out, NUT is reporting a 0% load on the unit while running my headless HP mini-pc server and it itself is set to gracefully shutdown at 10% UPS battery. So I pulled the cord from the wall and just let it do it's thing.
HOURS. It lasted frigging HOURS compared to the nonsense runtime numbers it was giving me. It sat at 24%/660 seconds for 45-60 minutes lol. Must be the recalibration.
It died at some point. I passed tf out before. Woke up in the morning and plugged the HP into a surge strip to bring back up those services, unplugged everything else from the UPC. It's currently charging no-load, reporting 67% to Grafana.
So, apparently one needs to do this process with every new UPS battery? I'm not sure if I'm being fed false info or not. I've never done that lol. Not like I'm working with UPS 24/7 but I've got 3 here at the house now and 2 of those have never given me any grief, and one of those is on it's third battery. Guess I'll keep it in mind for the future, it may have only been necessary because of the situation the 750XL came out of? IDK it's all weird.
I'm also confused as to which unit actually better suits the demand here. With the tiered shutdown of the server the load is closer to 60-80 watts, and I'm lead to believe the 850M is actually more appropriate for this, but that doesn't buy me any extra runtime.
It has a 12v 9Ah battery, versus the 24v 18Ah of the 750XL...but the XL will be less efficient with such a light load? Not sure how much that matters in the long run, and I know it isn't so simple as "double Ah double time!" But they sell those expansion boxes for the XL line, which adds 2x18Ah batteries to the system. Regardless of efficiency that'd have to run the system longer.
IDK I'm just stinking out loud at this point and not sure which direction to go. Those expansion boxes aren't cheap, $200~ without batteries and $600 with.
Blah blahlahsfd;lahsdf gotta think it over some more lol