How do people handle Ad-hoc/SQL Centric Requests?

I’m just curious if anyone has found a really successful model for visualizing data that you don’t necessarily want built into your data model. I find myself faced with a few options and wonder how others are dealing with a similar scenario.

Reasoning :
Sometimes I have user maintained data I want to write SQL directly over, or a non-joinable data-set that I need for analysis and want in Looker for continuity, but don’t necessarily want clouding up a project or model.

Options :

  1. Run the query through pure SQL and export into google sheets/excel for reporting - works, but I hate reporting outside of a BI tool
  2. Pay for a low-cost provider for ad-hoc SQL requests - I’m thinking something like Redash
  3. Create a separate project (Ad-hoc) with it’s own repository and use SQL runner to create quick models to deliver insights.

For emphasis, I’m bought in to data modeling, I agree on Looker and data management for 80% of what I need, but the other 20% that’s quick delivery ad-hoc non-modeled analysis I’d still like an easier solution than spending time making it fit into and cluttering my model vs. working fast to deliver something in a timely manner.

Thoughts appreciated!

I have a view called misc_views, that contains multiple view definitions inside it. (One file, so doesn’t clog the project too much, the new folder feature is probably better than this)

Whenever I need to share results from these misc views, I set ‘hidden: yes’ on the explore, this way it doesn’t show up on the front end, but is still accessible via URL.


I really like the one file, no clog, hidden explore idea. That sounds a bit easier than setting up a separate project. Thanks!

1 Like