For me, trickling powder has always caused all of my digital scales to drift frequently. Ironically, my cheap Amazon digital scale drifts less than my Dillion or Lyman scales did (and also less than the RCBS/PACT scales I used to use). I weigh my pan empty, right after doing a calibration. For the sake of this example, lets pretend that my pan weighs 128.12 grains. While loading, if I pick up my pan do dump the powder into a case, the scale should ready -128.12. If it reads something that's off by more than .02, I consider re-zeroing everything. If the value is off by more than .04, I automatically tare and reweigh the charge that I just dumped. I've also used a sharpie to write the weight of the calibration standard in grains on the standard itself so I can quickly set it in my powder pan and make sure nothing has changed.
I've also gotten away from trickling in the standard sense. I throw my charge from an RCPS Uniflow that's mounted to my bench in an RCBS stand (not the sheet metal piece the thrower comes with). I throw directly into my pan, and then place the pan on the scale and weigh it. I have a second pan with a narrow spout that works great for scooping up a few kernels out of the charge, or for trickling a few more kernels into the powder charge. After 20 rounds or so you'll have a pretty good idea of how much you need to add or remove to get the charge where you want it. If its within +/- .04 grains, that's PLENTY good enough. For decades, our scales only read to 0.1 grains meaning that the exact same charge could ready 25.3, 25.4, or 25.5 grains and still be considered accurate. I can load this way a LOT faster than I could using something like an RCBS chargemaster.