The first stage of a multi-stage rocket is much less weight-sensitive than upper stages. Saving some fuel for a landing would not have the payload penalty that an upper stage or space capsule using a rocket landing would. The fuel percentage would be far less than 50 percent, but I don't have any hard data on what it would actually be.

It would be almost empty at landing, and not weigh much. Weight penalties for a rocket-recoverable first stage (aside from the fuel) include the landing legs.

Parachutes weigh something too. A lot, for large stages.

One problem is trajectories. It's very difficult to work a stage-one burn trajectory to allow a ballistic landing at or near the launch site, although I've seen some interesting "dogleg" tricks (on paper) to solve this. Easiest solution is a downrange landing site. Unfortunately, almost everything downrange of any eastern launch site is ocean, and overland launch sites give regulators hives.

Another problem is aerodynamics of the "fall" before you light the rocket. Not hard to get an unrecoverable spin started, and some solutions involve a small "drogue" chute just for stabilization, not slowing.

An upper-stage or orbital recovery would be at velocities where re-entry heating would be an issue, although the high velocity also gives you a lot more margin on landing sites.