Decides what to build, for whom, and how to tell whether it worked.
Why this one is hard to judge
The discipline is barely five years old, so titles say very little about what someone has actually shipped.
What to ask for
Ask for a feature they shipped and what they cut from the original idea.
Ask how they decided a feature was ready to ship, given it can never be perfect.
Ask for a user complaint that changed the product, and what they changed.
Ask how they explain to users when the AI feature might be wrong.
The mistake most hirers make
Hirers choose someone who is good at describing AI possibilities but has never had to decide what a real user actually needs today. Product judgement here means knowing what to leave out, not what to add. A roadmap full of exciting features and no shipped decisions is a warning sign.
What good looks like after 90 days
A feature shipped that users actually use, with a clear account of what was deliberately left out. One change made in response to real user confusion or complaint. A working way of setting expectations with users about what the AI can get wrong.
How we assess it
Against a rubric that is published in full, on evidence the practitioner supplies and a reviewer checks. Where something has not been verified, the profile says so.