Thanks for great Article! I think you need to include larger datasets in the test. It could have completely different results. Also you should notice that stalling redshift is not so easy as scaling of Bigquery (completely nothing to do here).
So I think it’s great idea to the second part, start from current datasets and scale systems 2x, 5x etc and measure how long it takes to scale & how it affected queries speed & price. Also it’s very interesting how parallel work affects systems, I can predict that Bigquery will work with almost the same speed (until you will reach limits for on-demand systems) but what will happen with Snowflake and Redshift?