What I find interesting is that you used “SUM” and “COUNT” in the first pass at all. Being used to dealing with the lowest common denominator of data integrity in HIT, my first impulse is always to eyeball the raw data first. Saying

Select SUM(Amount)/COUNT(*) FROM [dbo].[Payment]

Makes a lot of assumptions about the quality of the Payment table which I have learned the hard way aren’t always true. I always start with

Select Amount FROM [dbo].[Payment]

then eyeball the data returned and count the rows. Then (baby steps)

Select COUNT(*) FROM [dbo].[Payment]

Still make sense? Then

Select SUM(Amount) FROM [dbo].[Payment]

Still looking copesthetic? Great, now I am ready to

Select SUM(Amount)/COUNT(*) FROM [dbo].[Payment]

Sad, but true. The lowest common denominator of data analysis is often times the best place to start. Particularly if your dealing years of poorly maintained data on a 20 year old legacy system.