A long pillar page wins when the topic is one coherent journey best read in a single place and its sub-points lack standalone search demand. Several small pages win when each sub-point carries its own distinct intent and demand that deserves a dedicated answer. The pivot is one-journey versus distinct-demands, and it cuts cleanly between the two poles. Neither “long pillars rank better” nor “split everything for long-tails” is a rule; both ignore the only thing that decides it, which is what searchers actually want.
The pillar wins when the material is genuinely one thing. If a reader needs to move through the whole topic in sequence to understand it, and the sub-sections would be thin, demand-starved pages on their own, then one comprehensive page serves the journey and consolidates relevance instead of scattering it across fragments that compete with each other. Breaking that into ten small URLs creates thin pages with no independent demand, which dilutes the topic rather than covering it better.
The cluster of small pages wins when the sub-points are not really one journey but separate questions people search independently. If each sub-topic has its own intent, its own searchers, and could each be the best answer to a distinct query, then dedicated pages let each one compete on its own terms with full focus. Forcing those distinct intents into one giant page makes it a generalist that ranks for nothing in particular, losing to focused competitors on every individual query.
So the decision rests on a single check you run before splitting: do the sub-points carry standalone search demand and distinct intent, or are they steps in one journey with no independent pull? If they have their own demand, build the cluster. If they are one coherent read with thin sub-points, build the pillar. Decide on demand and journey, not on a preference for big pages or for many small ones.