Correction sized to the error
Proportional control is the intuitive starting point, the correction applied is simply the error multiplied by a constant gain, usually written Kp. If a drone is 2 meters below its target altitude, a proportional controller pushes the motors harder than it would if the drone were only 10 centimeters below target, because the correction scales directly with how wrong things currently are. Turn Kp up and the system responds more aggressively and reaches the target faster, turn it down and the response is gentler and slower. This single number is often the first thing engineers tune, because its effect is the most immediate and easy to see.
The relationship is linear and has no memory, the controller doesn't care how long the error has persisted or whether it's growing or shrinking, it only cares about the error's size right now. That simplicity is also proportional control's biggest weakness, described below.