How big a territory should be

Size a territory by how many doors it holds and how many passes you want over the season, not by how it looks on a map. Too large and every address gets worked once at one hour, which leaves most of the value untouched. Too small and the rep exhausts it and starts drifting across boundaries.

The arithmetic

Doors a rep works per day, times days in the season, divided by the number of passes you want. That gives a door count, which you then convert into ground using the actual density of the neighbourhood rather than its apparent size.

Too large is the more common error

It feels generous and it produces single-pass coverage, which given how many doors end in nobody home means most households were never actually met. The territory looks worked and is not.

Too small has its own failure

A rep who genuinely exhausts an area starts working the edges of somebody else's, usually without deciding to. Undersized territory is one of the more common causes of accidental backdooring.

Common questions

How many doors should a territory have?
Enough for at least two passes over the season at the rep's realistic daily rate. The absolute number depends on density and season length.
Should territory size vary by rep?
By throughput, arguably yes, and it is politically difficult. Reps read a smaller territory as a demotion regardless of the reasoning.
What if the rep finishes early?
Re-work it at different hours. That is almost always higher yield than expanding into new ground.