Frankly, at Lyft, we had a very good system for dashboards.
Lyft uses mode [0] for most dashboards, it has a very well documented data catalog thanks to the system they created, Amundsen [1], and the tables are almost all free to peruse by anyone at the company.
This lets anyone with curiosity and the willingness to write some SQL build some very useful dashboards. And every dashboard is very easy to understand because the SQL can be inspected and the sources can be inspected. If you wanted to go further, you can even pin down where and from which repository the data got emitted. Mode has a very poor charting library, but it satisfies 80% of the needs and the other 20% can easily be complemented by the fact that they have an integration with Jupyter notebooks that gives you even more power.
The entire system felt very ergonomic. It also felt like it gave the opportunity for non-technical people to step up and get their hands dirty rather than wait for a data analyst.
Lyft uses mode [0] for most dashboards, it has a very well documented data catalog thanks to the system they created, Amundsen [1], and the tables are almost all free to peruse by anyone at the company.
This lets anyone with curiosity and the willingness to write some SQL build some very useful dashboards. And every dashboard is very easy to understand because the SQL can be inspected and the sources can be inspected. If you wanted to go further, you can even pin down where and from which repository the data got emitted. Mode has a very poor charting library, but it satisfies 80% of the needs and the other 20% can easily be complemented by the fact that they have an integration with Jupyter notebooks that gives you even more power.
The entire system felt very ergonomic. It also felt like it gave the opportunity for non-technical people to step up and get their hands dirty rather than wait for a data analyst.
[0] https://mode.com/ [1] https://www.amundsen.io/