This is an exercise (2.2:3) in An Invitation to General Algebra and Universal Constructions by George M. Bergman:
If $G$ is a group, let us define an operation $\delta_{G}$ on $|G|$ by $\delta_{G}(x, y) = xy^{-1}x$. Does the pair $G' = (|G|, \delta_{G})$ determine the group $(|G|, {}\cdot{} , ^{-1}, e)$? (I.e., if $G_1$ and $G_2$ yield the same pair, $G'_1 = G'_2$, must $G_1 = G_2$?)
($|G|$ means underlying set of a group).
If $G'_1 = G'_2$ then they are equal as pairs and that means that $|G_1| = |G_2|$ and $\delta_{G_1} = \delta_{G_2}$. Let $\circ$, $^{-1}$, $e_1$ (resp. $\star$ , $^{'}$, $e_2$) be binary operation, operation of taking inverse and identity operation in group $G_1$ (resp. $G_2$). By virtue of $\delta_{G_1} = \delta_{G_2}$ we get $$ x \circ y^{-1} \circ x = x \star y^{'} \star x \tag{1}$$ for every $x, y$ of $|G_1|$. Plugging various $x$ and $y$ in $(1)$ we can get following relations of identities: $$ e_1 \star e_1 = {e_2}^{-1} \tag{2}$$ $$ e_2 \circ e_2 = {e_1}^{'} \tag{3}$$ $$ {e_2}^{-1} \star e_1 = {e_2}^{-1} \circ {e_2}^{-1} \tag{4}$$ And I'm stuck with that: can't get any reasonable relation which would allow to get equalitity of identities or equality of inverses or just distributivity between two group operations (to prove that groups are equal). On the other hand, to disprove that one must find a counterexample e.g. in groups of small order (i.e. two different group structures on the same set giving the same derived operations), but it is tedious.