With cold material the fluctuations happen fairly quickly when measuring at the heater itself. I use both methods on my machines at work. Some heaters have the thermocouple built into itself and others are embedded into the heated mass.
The ones that aren't directly on the heater very often overheat during the lag behind turning on and getting the t/c up to temp. This way works faster but risks excess temps.
The ones built into the heater itself fluctuate quickly but never overheat what I'm heating.
There's a balance to be reached between the two extremes of course.
I don't know if you saw what I did. I have two thermostats, wired in series. The probe for one is on top of the aluminum plate below the cells. The heating pads are mounted below the plate. This thermostat - call it T1 - is set to 90°F with a hysteresis of 10°F. The probe for the second thermostat (T2) is on the top of the cells, and the thermostat is set to 50°F (60°F for my testing) with a hysteresis of 10°F.
So in operation, it should work like this.
If the temperature of the cells (as measured on the top) gets down to 40°F, T2 turns on. If the temperature of the plate is 80°F or lower, T1 is ON. Since both thermostats are on the heating pads are turned on.
If the aluminum plate gets up to 90°F before the cells are warm, T1 will turn off until the plate cools down to 80°F again. So the heating pads are turned off, even though the cells haven't gotten to 50°F.
T1 may cycle on and off several times, making sure the heating is happening low and slow.
Eventually the temperature of the cells reaches 50°F, and T2 turns off. The entire heating system will then remain off until the cells drop down to 40°F.
Although this required the expense of a second thermostat, I feel really confident that it will allow me to heat the cells but not cook the side next to the heating pads.