What a lot of difficulty there is in trying to talk about psychiatric diagnosis. We try to say one thing but can easily end up meaning something else.
I have often taken a position that defends the value of diagnosis in mental health. People have often refuted that position by citing the failings of the DSM project, as though diagnosis and the DSM were the same thing. For a while I tried to resist this by pointing out (over and again) that I was not necessarily referring to that complex and troubled manual but to something like "classification plus a probable explanatory story". The point has never stuck, and I have to face the possibility that some of the fault is mine.
Why persist in talking about diagnosis? Why not seek a word that doesn't alienate people? Diagnosis seems to suggest "knowing", which it isn't if we're honest.
Perhaps instead of diagnosis, we could talk of "characterisation". When someone understands a problem in a particular way they characterise it, describe it as having a particular nature. "Your avoidance of parties is a social phobia, exacerbated by your ongoing avoidance and we can expect forms of exposure therapy might help." or "Your mood changes are like those that have been called "Bipolar", and when people have taken this or that drug they have found them easier to live with."
This way of talking can resemble formulation (the first example) or it can resemble diagnosing (the second), but it isn't supposed to be more like one or the other. Formulation tends, in this debate, to mean the idiosyncratic and "intelligible"; the formation of "meaningful narratives" (Boyle and Johnstone, 2014). Diagnosis tends to emphasise the regularities across cases, with "intelligible" referring to explanation in terms of medical as well as psychological processes (Hayes and Bell, 2014).
The idea of characterisation is consistent with either of these approaches. You can characterise a problem as psychosocial, as medical, or as a combination of both. If what we are doing is characterising, then we can take seriously the idea that someone is unwell when their mood consigns them to their bed for a fortnight. We can benefit from pattern recognition (Characterising the problem as a depression), without appearing to commit ourselves to belief in an entity that we can't yet describe (the "underlying" illness).
Characterising is more than classifying (because it speaks to how you view the classification), but less than diagnosing. It is a bit like formulating, but without the assumption that the explicable processes take place at the level of meaning. It's a clunky term, and not one that can be expected to "catch on", but when I talk about the value of diagnosis, it is this I am trying to describe.