When I had had a paid account with unlimited tactics I would sometimes go on long benders and do hours of tactics in a day, sometimes hundreds of them in one session. I don't advocate for this way of solving tactics as it just leads to frustration and a drop in focus like you say.
In my opinion the best way to get better at tactics is to do both slow puzzle solving and quick solving. Quick puzzles like puzzle rush is for reinforcing your pattern recognition, normal slow tactics are for building the memory bank of patterns. Use slow tactics to push your ability to calculate deeper and keep track of all the variations. Try to spend the time needed to calculate all the variations and potential responses, to the best of your abilities, before you make the first move.
One thing that helped me a lot was that I stopped the habit of guessing a move to get the next puzzle. If you get one where you don't see any ideas at all - take a step back. Try to deconstruct everything in the position; what are both sides trying to do, what are the weak points, undefended pieces, pieces that have limited mobility, etc.
I would like to get 2300 in puzzles this year. My peak was 2181., but this was exceptional. Mostly I am happy to reach 2100.
Sometimes I do puzzles with a horrible attitude and drop a lot of points in few minutes. Sometimes I try the next day and have the impression the puzzles are really easy and get everything right. I am still struggling to keep focus when doing all puzzles. Often I see an idea and give a try without any attempt of calculating the lines, and this is always a bad idea, of course. I think I should spend several minutes to calculate the lines, but should I really do this when the solution is quite obvious? Ok, sometimes the obvious moves are the wrong ones, I know, but often the solution is really simple. How do you deal with this kind of problem? Do you always wait until you are really sure, never given a try before 5 minutes thinking? Thanks for your feedback.