I have the following equation which is simplified from another relatively long equation. $$ xe^{-x}= c (\frac{x}{c})^{-x} $$ Here Lambert W function doesn't help since you still get $x$ inside the Lambert W.
I know that I can use numerical methods to find the interception(s) of these two lines. But in general, I wonder at what point we can realize that there's no hope of finding the closed-form solution and we should stop looking for it.
Assume that both $x$ and $c$ are positive. So, as @user619894 commented $$x^{1+x}e^{-x}=c^{1+x} \quad \implies \quad c=x \,e^{-\frac{x}{x+1}}\tag 1$$
$$\frac{x}{x+1}=y\quad \implies \quad c=\frac{e^{-y} \,y}{1-y}\quad \implies \quad \color{red}{e^{-y}=c \,\frac {1-y}y}$$ which has an explicit solution in terms of the generalized Lambert function (have a look at equation $(4)$).
This being said, if it is to have obtained a closed form expression, from a practical point of view, it is not very useful. However, if we consider the case of large values of $x$ and $c$, we can write
$$x \,e^{-\frac{x}{x+1}}=\frac 1e \Big[x+1+\sum_{n=1}^\infty \frac {a_n}{(n+1)! \,x^n}\Big]$$ The coefficient form sequence $A111884$ in $OEIS$.
Truncated to some order and using series reversion (with $t=e\,c$) $$x=t-1+\frac{1}{2 t}+\frac{1}{3 t^2}-\frac{1}{8 t^3}-\frac{7}{15 t^4}-\frac{31}{144 t^5}+\frac{19}{35 t^6}+\frac{4927}{5760 t^7}-\frac{563}{2835 t^8}+O\left(\frac{1}{t^9}\right)$$
Using the above terms for $c=10$ gives the estimate $x=\color{red}{26.20165628306}409$ to be compared to the solution obtained by Newton method is $x=\color{red}{26.20165628306386}$
Even for $c=1$, the above truncated series gives $x=1.93317$ to bee compared to the solution $x=1.93295$.
Reinjecting this expression in $(1)$, the difference between lhs and rhs is $\frac{81479}{44800 e^{10} c^9} \sim \frac{8.25\times 10^{-5} } {c^9}$