It looks like your expression is big enough that it may exceed the buffer size inside WA.

I tried entering a slightly simplified version of your problem

solve s=(b/z+cz)/(a+b/z+cz+d), a=r+(iwl)/(iwm), d=a, c=1/(iwm) for s

WA page for this

and it seems like WA understood that.

Then I tried a second step where I inserted the value for b, that I left out of the first step, into the result provided by the first step.

s=((2iwm*(r+iw(l-m))+(r+iw(l-m))^2)/(iwm)mw-iz^2)/((2iwm*(r+iw(l-m))+(r+iw(l-m))^2)/(iwm)mw+2mrwz-iz^2)

WA page for this

It looks like it was just barely able to fit into the buffer and WA was able to understand that and gave a solution for s.

If in the second step I instead I tried "solve s=..." from the first step followed by "b=..." then again it didn't appear that it would fit and WA was not able to understand that.

For larger problems, trying to find a way to successfully break a large problem into smaller steps has been very useful in the past.

You just need, as always, to be very careful to check that the results are correct.