As both the photon and the electron are point particles in standard QFT, it would be basically undefined what "head on" means. Two points can't really hit "head on" since they have no spatial dimension.
There are different ways you can think about the problem, but two billiard balls hitting right head on is probably not a good way. In semi-classical quantum mechanics (Compton's original derivation uses a semi-classical approach, basically appealing to quantum mechanics only in postulating a photon), from which the photon is modeled as an object with momentum p=h/λ, one should probably just think of the two particles as colliding approximately head on, allowing for the photon to scatter at random angles and imparting an angle dependent amount of momentum to the electron.
Another way to think about it, supported by the Feynman diagram for this process, is to think about the electron actually absorbing the photon, and then re-emitting it in a different (random) direction. If this is easier for you to visualize, you can think of it this way too.
Really, quantum mechanics doesn't say much for the "moment of impact" since all the tests we can do are scattering events where we see the initial state (2 particles going in) and the final state (2 particles coming out) but we can't really ever observe the "moment of impact" itself.