Tactics Trainer as a tool for Improvement

So how do other people use the tactics trainer for improvement? Obviously it helps with pattern recognition. On chess.com I do the timed problems, so it is mainly exercising chess intuition. On chesstempo.com I do the non-timed where I am focused solely on analyzing until I find the solution (even though I sometimes revert to my timed habits from here). 

What I am trying to decide is how I should interpret the results of this work on here. As I made the recent jump from the 2300-2400 rating range into the 2400-2500 rating range, I have had a marked drop in success rate %, roughly 1%. That doesn't seem like much, but when you consider the number of problems that have been worked, ~23k, that actually is a noticeable drop in short term success rate. It would be natural I guess to hit a point, especially at first where success rate falls off. I still get enough correct that I keep the rating in the 2400s, so it is not like I am completely lost. What I am really trying to figure out is it is better to have a higher success rate and a lower rating, or a higher rating and a lower success rate. Of course a higher rating and higher success rate would be best. I'm not sure if it would even be possible to achieve the 1st case in the long term, because as your success rate increases your rating will naturally go up. This is especially true when you factor in the fact that you get partial credit for partially correct solutions.

Speaking of tactics problems, here is a nice problem that is part of the exam for the 1st lesson in Yusupov's Build Up Your Chess: The Fundamentals. I'm not sure that I have the Q-side completely right because I am recalling the position from memory and I haven't looked at it since this past weekend. I will look when I get home and update the diagram if it is incorrect.