A route you can defend beats a shorter route you cannot. On repeated days, the defensible route wins.
There is a difference between the shortest route and the best route, and the gap between them is where most improvement happens. The shortest route is a property of the puzzle. The best route is a property of the player.
A best route is one that can be explained before it is executed. If the reasoning at each rung can be said out loud, the route is reusable. If the route was found by trying words until one worked, it is not a plan; it is a lucky walk. On a single day the lucky walk may well be shorter. Across a month, the player with defensible routes always wins, because defensible routes teach pivots and pivots recur.
This is why move counts alone are a bad scoreboard early on. They reward guessing in a game that is mostly about structure. A better early scoreboard is the number of rungs that were predicted by the previous rung's reasoning.
The exercise is simple. After a solve, write the route as three sentences: why this pivot, why this corridor, why this ending. If any sentence is missing, the route was not planned. Then replay the same ladder a week later and see whether the same three sentences come back. Practice ladders are the place to do this without pressure, and the unlimited ladder can generate as many repetitions as the reasoning needs. Routes that survive a week of replay are the ones worth keeping.
The practical version of this idea is to keep a small collection of routes rather than a small collection of scores. A route can be replayed, questioned and improved. A score can only be compared, and comparison is the least useful thing a player can do with a month of play. Keep the routes, and let the scores fade.