The fit score had one last loose edge.
Dog pages, feed surfaces, and payoff moments were gated, but chat could still reveal more than the rest of the product allowed. If the assistant can see a number, it can say the number.
The free look
I kept one signed-out preview.
That felt important. A fit score is easier to believe after you have seen it once. The first score gives someone a real taste of the value. After that, the product can ask for sign-in without feeling like it hid the entire idea behind a wall.
The locked state
The chat card now has a proper locked version.
It can explain that there is a fit read available without leaking the actual score. After sign-in, the card can become actionable again. The locked state should feel like a continuation of the same experience, not a dead end.
The rule
The important part is that the rule lives where the value is created.
It is not enough to hide a number in the interface if another surface can still ask for it. The product has to decide what the assistant is allowed to know in that moment.