Question everything, especially what you think is right. This is actually putting you on the right path.
You make a list of things that don't "feel" right. Then you explore resources like this message board on how to do it right. A perfect example is tabled layout. When is it justified? Am I causing problems in accessibility, does it affect the load time of my pages, just how important is the framework of my pages?
You read, read read; listen to what others say, take the pains to try it. At least try it. You may fail at first. You try again, and again, until it is no longer a topic you fear, but one you know quite well and can address with a degree of authority. Then incorporate it into future projects and pages.
So now you have a new rule in respect to how and when you use tables, and you stick to it. You may waste a whole day (or several days) gaining insight to the problem at hand, but you've actually done something very valuable: you've learned the right way to approach a problem. And, if you stick to tables for layout, you will have valid arguments as to WHY you decided to do so.
Cross one item off your list, and go to the next one.
Eventually you'll see that the world gets bigger and bigger, and the more you know, the less it seems you really do know. Which leads to more items to add to the bottom of your list.
In the meantime, your pages get better: better indexing, more accessible, fewer and fewer things begin to "feel wrong."