This is actually a bit of a tough question, and one that probably has many correct answers depending on the skills of who is asking it. While I fully agree that actually sitting down at the bench and putting the time in is more than likely the best way to improve, there are those who do put the time in but always take the same exact steps and go through the same exact motions each and every time. With that in mind, there will obviously be no 'improvement' if the builder never tries anything new or alters any part of their routine.

IMO the best way to improve is to try new things, or in many cases try to fix or work around something you may have 'done wrong'. Far too often I read "oh this happened so I scrapped it/put it back in the box/called it close enough/smashed it etc.", which teaches us absolutely nothing about improvement, learning to work around our mishap or correct our mistake. We risk doing the same exact thing on the next model and if we didn't teach ourselves how to work with it, we will simply repeat the same action and give up/shelve/destroy it. The best lessons I have ever learned in modeling over the years have come from pushing forward through that goof that I thought ruined the model and making it work in the end, it doesn't always work but at least seven times out of ten when it does, you experience a growth spurt in your skill set that will push you to another level of the craft.

The second best way to improve is to listen to and heed the advice of accomplished modelers. All to often I see great advice ignored or challenged, as if the person asking the advice is more interested in a debate, or proving that his method is just as good/better than he is in learning a skill, technique or method. Now I don't mean the guy with ten accounts on ten different forums and tens of thousands of posts talking about modeling, I mean the guy whose name you may recognize from his many completed models in different forms of media and the like. If someone whose body of work that you can actually see and study and decide that he has what it takes or not gives you advice, I'd recommend that you digest it very well and put it into practice. If you are fortunate enough to have the ear of someone whose work you admire, capitalize on that and keep the dialog going and practice what you have been told. By practice, I mean try it more than once and not simply shrugging it off as not working for you if you don't get stellar results the first try. Put in the time to figure out why it isn't working with you, experiment with it, ask more questions of that modeler. Don't just throw your hands up and give in, re-read the advice/article/thread etc. and work out what/where it went wrong.

These days it seems to me that most every modeler is on fast forward trying to learn all the best tricks and techniques before nailing down any of the fundamental basics. So many rush out to buy the latest book and newest paint set and most expensive airbrush thinking they now have the magic tools and simple steps in creating the same artwork found in that book, all the while disregarding the ten-twenty odd years of practice and learning that modeler put in before he was accomplished enough to write that book.