If factor isn't found in P-1 algorithm, should upper bound be increased linearly (i.e. +1)

I have seen some implementations of Pollard’s P-1 algorithm where the upper bound is only increased by 1 if no factor is found. Such an implementation is described here. Is it sort of missing the point in terms of efficiency to only increase by 1, or is is it sill better than trial division for finding a factor? Is what they’re implementing in the link really the P-1 algorithm? I’ve seen other sources make it sound that a certain amount of intuition is needed when choosing parameters and you shouldn’t just increase by 1 each time. Am I understanding correctly?