HubSpot caps lead scores at five per account. Not five per object, five in total. Most teams find this out after they have already spent three of them, and the third one is usually persona matching.
Persona matching does not scale as a score. You have four personas today and six next year, and a score can only ever produce one number. So the score becomes a proxy for one persona, or a blended figure that means nothing, and either way you cannot answer the question people actually ask: why was this contact tagged as that persona?
It also competes for a slot with things that genuinely need one.
One calculated number property per persona on the Contact, named {persona}_match_strength, computed by workflow logic from form answers, behavior and source.
A workflow that picks the winner. It reads every *_match_strength, writes the highest into hs_persona, and writes a persona_confidence_label alongside it: Strong, Moderate, Weak, or Unable to determine, based on the margin of the win and the absolute score.
Two things fall out of this that a score cannot give you.
It is debuggable. Every per-persona property is individually inspectable, so "why is this contact a Founder?" is answered by reading four numbers rather than by guessing at a black box.
It carries its own uncertainty. A contact who scores 40 and 38 against two personas is not the same as one who scores 90 and 12, and the confidence label says so. A single winning enum hides that completely, and the hiding is where bad routing comes from.
It also scales to any number of personas, because you are adding a property rather than competing for a slot.
Reserve them for things that earn them. A working set:
The rule that keeps this honest: if you cannot justify spending one of five on it, do not build it as a score. That question gets asked out loud, and it kills about half the proposals that reach it.
A platform limit that is low and account-wide is not an annoyance to route around. It is a forcing function. Five slots means five things matter enough to be scored, and having to name them is more valuable than the scores usually are.
Most accounts we look at have scores nobody remembers building, still firing, feeding nothing.