In theory, an idle descent from cruise altitude all the way to the ground would be most efficient. I believe it was actually common to at least attempt this feat in the early days of jet aircraft, because they burned more fuel even at idle and there wasn't nearly as much traffic to contend with. Obviously much less common in today's ATC environment, and with modern turbofan engines you don't want to be at idle on approach because of the spool up time in case of a go around or other issue.
Nowadays everything is managed by the FMC. I can only speak for my airline and the 737, but the dispatcher computes the flight plan based on current conditions and manages the inputs to achieve the desired outcome. That outcome can be best economy (lowest fuel burn), best time (shortest flight time) or anywhere in between. The software then outputs a Cost Index which is transmitted to the airplane with the takeoff data. The FMC uses that CI and the current conditions (weight, temperature, winds, etc) to determine climb/cruise/descent speeds.
The FMC in the 737 calculates an idle descent profile to comply with any airspeed/altitude restrictions in the flight plan. It will fly the speed (Mach to IAS) that was produced from the CI inputted. At least here in the USA almost every commercial airport is served by an RNAV arrival with multiple altitude/airspeed "gates" at fixes along the arrival. The FMC still plans for an idle path descent, which normally results in level offs after a fix until the next idle path is reached. When ATC issues a descent and you haven't reached the computed path it will descend at 1000fpm until intercepting the path or you reach the assigned altitude.
My airline famously flew every leg at the "barber pole" back in the days when gas was cheaper than water. That hasn't been the case for many years now and I assume it is also true universally in aviation. Fuel savings are a huge part of day to day operations both in the air and on the ground.