Even if it is a physical constant, $k$ may be the unknown - that'd be an inverse problem. What is known and not known should not change the model itself, so the logical order is to introduce all variables and discuss their physical meaning and relationship, and then before studying the mathematical problem, say what is data and what is unknown for your present work.
In general, this can take the form of stating your problem as
Find $(F,\Delta l) \in \mathbb{R}^2$, such that
$$F=k\Delta l$$
and
...
So, to come back to your question, it is usually "the other way around": we list the variables which are not known constants.
Some of the data may be also explicitly referred to in that statement of the problem, especially when there are restrictions on them (e.g., "Find..., such that... with $k \in \mathbb{R}^{+*}$")