finding deviations from the mean

cool2021, your template tool is really great and can be really useful for particular situations, I just don't know if this is one of them. I don't know what you mean by "more robust results". If by that you mean it will show more outlier data points then maybe but I would contend that would be "less robust results". I don't think they want to be distracted by needing to investigate every time a cloud hovered over 1 panel 1 day. The whole point of averaging and filtering and many other data techniques is try and parse out relevant and important information from the noise and the not important information. If your tool is highlighting all the noise and the actual bad data together then is that really "more robust"? To be honest I'm just looking at your output and saying to myself, If I was them, how could I use that output to do what I think they want and I'm just not seeing it helping. In the file I gave above, I averaged the output for each sensor across all the days for each giving timeslot to help filter out the nuances. Then I both highlighted the remaining potential outliers using conditional formatting and did a count of potential outliers for each panel; each based on 2 different thresholds (I used 1sd and 2sd, which is likely not the optimal but as good as any without more data/analysis). So I see this as a tool to help indicate which panels have repeatedly low output throughout the day (e.g. the panel is going bad or obstructed or dirty) and which panels tend to have low output at particular times during the day (e.g. the tracking mechanism isn't optimal or maybe location is shaded by something).
I really don't see what your analysis is doing for them. Also since it is column based I really don't understand what it is going to help because it has a natural trend from low to high back to low. And specifically in this above attached comparison you have 2 "outliers" highlighted and I look at the data and don't see what it is pointing out. That said I would have been really curious to see if you had extended the data set a few more rows to include the 9:00 hour to see what would happen. I'm guessing it may highlight a row there because the 9:15 data seems to have a drop for many of the panels (probably a large cloud). So PLEASE help me understand how to interpret your outputs so i can know when to recommend this useful tool.