Recognising yourself in a group of four is a different and easier task than choosing between sixteen. This asks you to do it three times: your temperament, your club and your quadra. Then it tells you which type all three point at, or, if they do not agree, asks one or two more questions to separate the candidates.
Two of the three would be enough to name a type on their own. The third is there to disagree with them if it is going to, because a check that cannot fail is not a check. If the test has already given you a type, this is the cross-reference it sends you here for.
Why small groups
A type is a structure of eight functions, and almost nobody can hold sixteen of those in mind at once well enough to pick between them. The small groups exist because each of them collapses that problem to a choice between four, on one dimension at a time, and each of the four is a description a person can actually recognise or fail to recognise in themselves.
The three used here are the three that work best for narrowing. Temperament is the most externally visible of them (how a person paces themselves, whether they push or adapt) and is often the one other people would answer most accurately about you. Club is about what you attend to without being asked. Quadra is about which set of values makes a room easy to be in, and is the one most often answered aspirationally rather than accurately, which is part of why it is asked last.
None of this is a substitute for reading the type profiles. What it does is get you to a profile worth reading, with a reason for having chosen it, instead of working down a list of sixteen. Where it lands on two candidates rather than one it says so, and the disambiguation tool takes a specific pair and asks the questions that separate those two.
When the answers disagree
If the three answers do not meet on one type, or you mark a question as a choice between two, the finder asks up to two extra questions. Each comes from another small-group system (cognitive styles, communication styles, stimulus groups, argumentation styles or romance styles), picked because it splits the types still in contention. The options are shown as A to D with no group names, because a name like "Aggressor" says more about how an option sounds than whether it fits. The result then says how confident it can be: consistent when all three core answers agreed, leaning when the extra questions favoured one type, and unresolved when they did not.
What this cannot do
It cannot see your function stack, and it does not try to. Three recognitions are three recognitions: they are worth more than one, and worth considerably more when they come from groups that cut the sixteen types in different directions, but a person who answers all three from an idea of themselves rather than from their behaviour will get a consistent and wrong result. That is true of every self-report instrument including this site's own test. The check against it is the same one as always: read the full profile of whatever it lands on, and pay attention to the parts you did not expect rather than the parts you did.