Sigma Computing

Makes 'easy' queries very easy to work on and see results visually immediately. Natural way to think about grouping objects but still showing the 'row' level data is amazing and a huge selling point on getting non data people to use the sheets shared with them.
Makes the mental structuring of large undertakings choppable into smaller to-do's and then joining sheets together

Expanding on my point in my pro's about chunking down large undertakings to smaller conceptual efforts per sheet.... this is not very performant. I have two projects where I create a base sheet and then have 2-6 sheets built off of it before potentially joining them back together to combine the aspects I've added. This however, results in the base sheet/query being run numerous times and is often times a time consuming query to begin with. In a code based environment this could be made more performant by using a WITH statement so it only has to execute once and can then be used multiple times

Joins have to be exact matches (no date comparison) and can only be on original columns from the data source. Therefore, if a column must be switched from integer to string, you have to create a reference to the sheet before being able to join to other tables.

The ability to connect to any data number of data sources across our enterprise data warehouse to extract useful insights for my team. Without a line of coding I can report, analyze, and visualize data from our CRM, IoT data streams, 3rd party APIs, and now .csv uploads. The full access is balanced by perfect row-level security controls so I can use the same system to power data dashboards and portals for internal teams as I do for external clients.

The upside of being able to access ALL data streams leads to a downside -- it can take a long time to process and load information. However I look forward to data materialization features in the works which will eliminate this problem.