Generally there is a tradeoff between speed and torque. To get high torque, a gearbox is used so the motor runs at really high speeds but moves the output shaft slowly with higher torque. If you want to track something in LEO, you need rather high speeds. If you've designed for high torque (either to get high holding torque or because you are moving a large dish), you'll need the motor to go _really_ fast which they just can't do. The system I built last year for work topped out at 20 deg/s on the azimuth axis but had 500 lb-ft of torque.
There's another tradeoff between speed and accuracy. The same gear reduction used to build up torque also provides lots of motor resolution. Typically, you want the encoder to be the limiting factor rather than the motor step size. The system referenced above had 0.01 deg encoder accuracy and much finer step size.
Axis configuration is another concern. Elevation over azimuth assemblies can't track directly overhead. Azimuth over elevation can't track at two points on the horizon. If you're tracking satellites, this usually isn't a problem as the orbits rarely pass directly overhead and tracking near the horizon has too much signal degradation. Tracking aircraft is a different matter.