Great minds think alike. I started implementing a similar approach a few months ago. In my case the reason for this deep dive was trivial. One of our portfolio companies had good overall numbers but a very long payback period. As aggregated numbers can hide good and bad signals I wanted to analyze the numbers on a customer by customer basis.

To do that we got the raw transactional data from the company and I built a similar spreadsheet for their ~700 customers at the time. This allowed me to see profit/loss for each customer. It’s amazing what you can get out of these numbers especially at pre product-market fit companies.