Drift in the reading changing over time with a known weight/tare that's not zero. I have a check weight I use to verify the scale calibration prior to use. I use it on both my GemPro 250 and CM1500. If one is off, the zero is done again and checked. If still off they get calibrated using calibration weights till they read the same. With the auto zero feature in the CM you can not do this at or around zero. I use a common charge weight in the 30gr range. Then just plot over time, and checking zero at the same time. On the CM it will start throwing charges heavy normally if a drift accours. When I detect this I will do a manual zero. Then check again. The reason I started re-zeroing every 10 throws. If in dought I put my check weight on to check. Normally I have found it it's not reading as it should the zero has shifted.
Since I had my CM apart cleaning it due to powder spill, I discovered those to screws holding the tray holder on played a big factor in the load cell. They have to be torqued to a value the will not allow yield during use. If light torque is used, zero drift is high. But once the torque is high enough it pretty much stopped all major drift in mine. Since I've had mine apart the calibration seams to be holding very well now. I've only had to calibrate a couple of time in the last couple of years.
I've learned over the years to always keep the pan/tray on the scale when waiting for the next charge. The hysteresis in the load sell will drift if not at a the normal state. This allows confirmation of Zero, just before you add the weight.
I've noticed the GP is more sensitive to zero shift if you do not keep the pan on. Keeping the pan on it's very stable.
In the old days most strain gauge load cells you stayed away from the bottom and top 10% of the scale for stability. With today's advanced electronics this can be controlled. Weather it can be controlled on the cheaper scales, depends on the mfg.
Thanks, now I don't have to open mine up again. 
