Yes from your description it sounds like there's no one driving the ship of your training, and you're getting disconnected haphazard (and contradictory) advice.
First, in the way of bad advice, to address one piece from this thread: it was mentioned that the ailerons' job is to keep the wing level, and this is not true. Their job is to maintain your straight-ahead track down the centerline, which may or may not mean keeping the wings level! Depending on crosswind (even tiny amounts, which change second to second) this will require small aileron inputs, resulting in barely-visible bank changes. Keeping the path straight with the ailerons, allows the rudder to do its job (which is completely separate and independent) of keeping the nose pointed straight (not the path, only the pointing).
If you do the incorrect thing and keep the wings level using the ailerons, this leaves both jobs of path and pointing to the rudder, which is often an impossible task to do both, and you have to choose one or the other. So the right way is ailerons to control path, and rudder to do the pointing. Now both tasks can be accomplished simultaneously.
Now for your landing question. Of course it's impossible to really see, but what I'm about to describe is a common issue, and the contradictory advice you quoted ("do not lift enough"/"lift too much") points to it. A lot of people are never taught what to actually look for/target in their flare, and are left to shoot in the blind with some sort of scripted sequence of increasing pull force, starting from a hopefully repeatable initial condition (flare height/speed, etc.) But if the initial condition varies even a tiny bit (say, arrived at the flare with a bit higher speed) or one of a number of other things varies later during the flare (catch a gust, or accidentally twitch your hand and pull more than you intended, etc.) that scripted pull sequence results in a balloon. Or, if the changes were the other way, you land too hard/early. How to react to that? If you were never taught what to target, and just do the monkey see/monkey do repetition thing, you've got no clue on how to improve. If you're observant and you think "I ballooned, I should flare less next time" that's a good mental attitude in observing the error and applying a correction for next time... but it's a naive reaction in this case, because the error was a result of a specific combination of the multitude of factors that was only in effect for THAT particular landing. And maybe in your next flare, you won't have that combination, or you'll have the opposite combination (you'll have too little speed where the last one had too much, or you'll catch a down gust where the last flare had an up gust.)
Another way to think of it is like this: Driving to work every day, do you stay on the road by memorizing every turn and turning the steering wheel by a scripted sequence of turns to match, which can be accomplished blindfolded if successful? Of course not, but the elevator equivalent to this, is too often the thing that pilots end up doing by being taught to "pull more/pull less" with no target reference.
So if you can't react to the last landing, and if the initial conditions don't allow for enough repeatability to do a scripted pull, then what do we do? Track a target! Namely, your height above the ground, which should always be decreasing and never increasing. And in the last few seconds, you can simplify away "not decreasing" to simply "flying constant height," (which does not mean constant attitude!) and it will come down and touch the ground anyway. You have to be constantly conscious of what's happening to your height, and constantly be making QUICK but SMALL corrections to what it's doing. If you start going down? Increase the pull. Stop going down at all? Or even worse, going up? Decrease the pull! These evaluations and corrections should be happening at least a few times per second. It may seem obvious, but it's not. So many times, the airplane will start ballooning up and up, and the student is oblivious and continues pulling, because that pull increase is part of their scripted sequence. It will never work. You have to be constantly and immediately RE-active to all these changes. To be clear, yes you should also try to give yourself the most repeatable starting conditions, but that, given all the possible following variations and upsets, will not nearly be enough. So you're not using a scripted sequence of turns on your steering wheel to match the road, but rather you're constantly watching how your actual position is doing compared to the desired position, and correcting accordingly.
Lastly, the "Rule #1" from a few posts up, which is woefully incomplete. It is BOTH a reading exercise and a brain-eye-hand-feet coordination exercise. If you're just out there shooting in the blind hoping for lucky results, and then trying to repeat those few diamonds in the rough, it's hopeless. You have to understand the concepts behind WHAT you're tracking and HOW you're tracking it, and internalize those into your brain (under the comfort and lack of competing tasks, of being outside the cockpit) for you to have a chance to then engage those concepts under the physical reality of actually doing it.