Balasubramanian Murugesan , for 1) see https://docs.microsoft.com/en-us/mem/analytics/data-collection . 2) i don't quite understand the question, but you should think about a baseline in terms of a snapshot of your metrics frozen in time. For example, say you create a baseline before replacing a bunch of old PCs to latest generation HW. You can compare the old scores for recommended software and boot performance before and after that change to have an idea of the positive impact on performance. Then say you define a regression threshold of 3% and take a snapshot before installing a third party AV. Your boot scores regress by 5% due to added time from loading additional agents, then your boot performance metrics will show as red because the regression was higher than the threshold you define . and 3) We feel the docs provide a good summary and there's no need to document the precise details, ie: The Software adoption score is a number between 0 and 100. The score represents a weighted average of the percent of devices that have deployed various recommended software. The current weighting is higher for Windows than for the other metrics since users interact with them more often.