L4E should be 2-looks for the entire solve, at most.

Look 1 (done in inspection): Blockbuild a V. If you can't see your entire V in inspection, your solutions are most likely quite inefficient. Instead of solving the centers and then 2 edges, try to solve them simultaneously. Generally in inspection, I follow this procedure: look for any edges connected to their respective centers, and then plan out how to blockbuild a V with that edge as a starting point. If there are no already "solved" edges, there's almost always a way to do one move to solve one, and then proceed to make a V with that one.

Look 2 (try to do this in inspection, otherwise look ahead while blockbuilding your V): Intuitive L4E. You should know all the cases in a single look, even if you go about them by solving one edge and then doing L3E. As far as looking ahead into L4E, here's what I do: look at the unsolved piece on your V-layer. The color on the V-layer should be the center you put in front on the top layer when you start L4E. (e.g. If you have a blue-yellow edge as the unsolved piece, with the blue sticker facing down when you start L4E, for most cases you'll have to turn the top layer so the blue center is facing you. Then you'll be able to solve the blue-yellow edge and reduce to an L3E case.)

Another major problem most people have at some point is solving tips efficiently. Whatever you do, DO NOT save more than 1 tip (or 2 in very rare cases) for the end. You should solve the tips while you solve other pieces, so they add the least amount of time to your solve as possible.

If you do all this well, there should be 1 pause, maximum, in your solves. If you can solve a V in 3 seconds, pause for 1 to recognize L4E, and do L4E in 3 seconds, you're already averaging 7, with quite a few clear places for improvement.