After "calibration phase" good regulation. Window open detection deficient.

When the controller and the room temperature (measured separately) observed a few days and then re-adjusted the temperature and the switching times accordingly, you will achieve a reliable and constant room temperature.
The temperatures indicated are in no way to be held as the room temperature to see - me be displayed 20.5 and 21.5 that brings real temperature. But that principle does not change - only to pay attention to stop it.
The volume of the variable speed motor is "clearly audible" but not "disturbingly loud". In the bedroom, it might be disturbing to sensitive fellow. (As a guide, perhaps as loud as the engine of a CD tray on a PC or DVD player - only the motor does not stop the noise).

What is really bad dissolved or perhaps theoretically not better solved was: The "Window-Open-Detection".
What still works quite well with strong freezing temperatures in the dead of winter (since large temperature differences exist), then it works no longer in mild plus degrees. The temperature differences are probably too low as the regulator could see the open window. The difference threshold or sensitivity can not be adjusted, unfortunately.
Thus, one must unfortunately but the regulator runterregeln when you open the window 'just in' Lock 'mode bit tedious. The regulator of Conrad with the external sensor should there be better, according to Stiftung Warentest. But they are also more expensive. But I do not know which, so I can not confirm this here.