A team map, not a scoreboard
Start by having each person take the free MBTI test on this site individually, answering as they usually are rather than as they wish they were or as they think a good team member should be. Nobody answers for anyone else, and nobody is required to share a result they would rather keep private. A map built on honest answers is useful; a map built on wished-for answers describes an imaginary team.
Once the results are in, lay the types out side by side. For a small team, the clearest way to read the map is by dimension rather than by all 16 types: how many people lean toward extraversion and how many toward introversion, how many toward sensing and how many toward intuition, how many toward thinking and how many toward feeling, and how many toward judging and how many toward perceiving. A dimension view shows the shape of the team at a glance, where a list of 16 type names mostly shows a list of names.
The purpose of the map is to understand working styles, not to rank people. There is no best type for a team, and a team of 16 identical types would be a worse team, not a better one — it would share the same strengths and the same blind spots. Treat the map as a conversation tool: something the team looks at together and talks about, and something worth revisiting when the team changes.