We use it as our infrastructure for big-data analysis and we get the results at the speed of light.
Every 60 seconds, there are nearly 1K queries that analyze users and IPs behavior.
Our entire ML is based on it, and there are half a dozen of applications in our platform that make use of BQ - from detecting scrapers, analyzing user input, defeating DDoS and more.
The built-in (with legacy sql variant ) Math and Window functions are super handy, the easy DataLab  integration, and the last but not least, Google DataStudio that let us generate interactive reports in literally minutes are all making our choice (3+ years a great one).
BQ replaced 2,160,000 monthly cores (250 instances * 12 cores each always on) and petabytes of storage and cost us several grands per month.
This perhaps one of the greatest hidden gems available today in the cloud sphere and I recommend everyone to give it a try.
A very good friend of mine replaced MixPanel with BigQuery and saved nearly a quarter of a million dollars a year since .