This site argues that a single number per trait hides more than it shows. It also gives you a five-letter type code.
I want to defend that, because it looks like a contradiction and very nearly is one.
Why build one at all
The Big Five has been correct and unread for forty years. Myers-Briggs is not a better model of personality — it is a worse one — and it wins comprehensively on the only thing that matters for reach: people remember their four letters, tell other people, and come back.
An instrument nobody reads has no effect on anybody's self-understanding, however well calibrated it is. So: ten pole letters, one per pole, with both ends of every axis written to be worth being.
The fix the other systems cannot make
A letter is assigned only when the confidence interval clears the midpoint.
That is the whole difference. If your Openness sits at the 52nd percentile with a band from the 34th to the 71st, you do not get a letter for it — because the honest statement is that the instrument does not know which side you are on, and a system that hands you a letter anyway is manufacturing a distinction out of noise.
The criticism that actually bites against type systems is not that they have types. It is that roughly half of people get a different code on a retest a few weeks later, and the instrument has no way of having warned them. Every other test hands out crisp letters and says nothing about how firm any of them is.
So this one computes it. My own code holds all five letters about 41% of the time. One of my letters is a near coin toss — the page says so, in those words, on my own result.
Publishing that number is the difference between a type system that admits what it is and one that does not. It is also, reliably, humbling.
Sitting in the middle is a result
There is a thirty-third outcome for people whose letters are mostly undecided. It is not an error state and it is not a consolation prize: sitting near the middle on several axes is a real way to be, and it is arguably more common than any of the corners.
Getting the condition right took two attempts. The first version withheld letters for the wrong reason and told somebody with a 95th, a 78th and a 10th percentile that most of their traits sat near the middle. Whether an interval clears the midpoint is a statement about precision; whether the facets inside a domain agree is a statement about heterogeneity. A domain can be precisely measured and internally varied at the same time — and both facts are worth telling somebody, which you do by saying both, not by suppressing one.
The flattery I had to design out
There is a figure on the results page for how distinctive a profile is, and the obvious way to compute it is distance from the average person.
Do that and almost everybody comes out unusual. Over five independent traits the typical distance from the origin is already substantial — measuring from the middle tells more or less the whole population that they are far from it.
That is flattery, and this project exists specifically not to hand it out. So distinctiveness is measured against the distribution of everybody's distances, and about half of people are told their profile is fairly ordinary in magnitude. Which is what "ordinary" has to mean if the word is to mean anything.