As 'cbarn24050' says, it depends on the application. A motor running at 70% load is quite typical in industry. The big question is, what is going to be your return on investment. Very little I would say. The efficiency of the controller is typically 97% and so you are going to lose 3% in real power if you keep the controller running all the time. Most people will by-pass a 'softstarter' once it has started to ensure they are not actually wasting energy during running. Look at the application more carefully and see what is going on. It might be more prudent to just switch the motors off if they are not doing much. There's a real reluctance to switch motors on and off due to the high starting currents, but if you are looking into softstarters, then switching the things off and on when needed (so long as it is not too regular) will save the most in energy.
If you want to calculate the 'savings' from your PowerBoss, we need to look at your motors (22 to 75kW). Using the 75kW as one example,(and if I use the data from a Siemens motor I have handy)the efficiency of the motor is 94.3%. This means that at full load the input power required to run this motor will be 75/0.943=79.53kW. The maximum losses are (79.53kW-75kW)=4.53kW. As mentioned earlier, the optimising function of a softstart will recover some of the fixed exitation losses, and these are typically 35% of a motors TOTAL losses=(4.53*0.35)=1.58kW. The PowerBoss that you are about to purchase will, they may claim, save 35%! This is 35% of your fixed exitation losses. Now we are getting closer to the real truth; 1.58 * 0.35= 0.553kW. This is it, at full load. So, now multiply the savings in kW (0.553) by your yearly running hours (say 8000 if running most of the time) by the tariff charge (no idea where you are but I'll guess the USA, so 10c/kWhr) and you get a typical yearly saving of $442 per annum. This is, by the way, assuming a controller at 100% efficiency. Take off 3% from your kW figures and you are somewhere close. You can work out the same at 3/4 load, assuming the efficiency of the motor is about the same.
Once you've done this, work out your Return and see if the sums add up.Take into consideration the running hours and I would doubt it.