What’s something beginners focus on that barely matters in real work?

Feels like a lot of early learning is centered around things that don’t show up much day to day. Stuff like squeezing out tiny model improvements, memorizing algorithms, or obsessing over which model to use.

But in actual work, it often comes down to messy data, unclear requirements, and getting something usable out the door.

Curious what others would put in this category. What do beginners over-index on that ends up not mattering much?