One day? Get them to stay one hour and you've made a huge improvement.
Hitech posted stats one time. I don't have the link handy, but from memory that out of the thousands and thousands that downloaded the trial during the Steam launch, the significant majority didn't stay 30 min and most never got off the runway.
New user training? Maybe. I'm not convinced. Sure, the intricacies of ACM take a life time of study, but just getting wheels up is not that big a deal. It's not massively different that IL2, or DCS, or any number of flight sims. But HTC does a pretty good job with the default control mapping that just about any $10 twisty stick works fine first try. At least enough to get wheels up. Would necessarily hurt to have a 5 min first flight automated walk thru. I don't know if that would turn the tide.
I suspect a percentage of them have already decided the game isn't for them when they launch and get a message that it isn't a game with a permanent F2P plane-set and more stuff if you subscribe like they are used to from something like War Thunder (with 70 million registered users they have set the market expectations on how a subscription service should monetize).
Instead they get a dialog right a start up that, nope, two week trial and then you can only go into some empty unused arenas. That is not what they were probably expecting. So they already know its not worth an hour of their time to try.
HTC has stated a small F2P plane-set is not an acceptable approach, so I don't know where you go from here.
All of the players that I've brought with me in the last few days are younger WT players and long term ones at that, and every single one of them wants to sub asap. I can tell you from their raw reactions that the biggest issue isn't the concept of the game (they quite literally feel that AH is everything WT Sim should be) it's the very, very clunky UI that causes them massive amounts of issues. They desperately love what I've put on the table for them, but they also admit the UI would have scared them off without me. So far my success rate of handholding them in is 5 out of 6. The only one that I've lost was due to the way the game handles higher ping (due to them being UK based) around the time of gunshots.
Do not discount how absolutely fed up a massive percentage of the WT community is with the way the game is being managed, AH offers lightning in a bottle to them if they can only get passed the UI clunkery, and once they know it they want to latch onto it like drowning rats. Almost everyone that plays WT wants a better game, AH can be that better game even if its not as 'pretty.'
To your statement about most not getting off the ground..air starts for newbies 2 week trial?
This is a non-issue, I won't argue semantics about flight models with anyone but AHs flight models are toylike (especially with the combat trim system) compared to Il-2 and DCS. This isn't an attack, it's just fact from a player side. Especially for players that are used to those other games FMs, even WTs 'feels' more responsive but that's NOT the issue. It doesn't matter if those games are 'realistic' or not, the complexity is just much less here and I'd argue that this is on purpose. That is a strength if you double down the concept of being a rabidly vicious BFM sim.
Il-2 and DCS offer dozens and dozens of failure points where you, as a player, can interrupt your own combat experience by inexperience without ever even engaging in a fight. Aces Highs assumption that the pilot in the aircraft knows what he's doing systems wise is smart, and you should focus on that concept. Which isn't to say 'we're an airquake hyper bfm game' but AH makes the player feel empowered in their actions without a need to go read dozens of pages out of a manual, or watch an hour and a half of YT videos just to handle basic flight.
You can, and I have, gotten 5 players almost entirely on the up-and-up for basic gameplay in less than two hours. That is a huge difference to Il-2s hours of engine management practice many feel they need, or DCS weeks of system handling. The issue is its handheld entirely, without what was essentially a UI interpreter they absolutely would have never given it a chance.