10 best practises to improve your Looker instance

Aurélien Roblin
7 min readMay 23, 2022

Hello everyone ! 👋

For 12 months I have been working exclusively on Looker. It’s a great BI tool, I love it !

It gives the ability to develop dashboards quickly and implement KPIs without waiting for the data engineering team.

It is very good !

But you have to be careful, your looker instance can quickly get messy.

In this article, I will share with you 10 of the best practices that we are implementing in my team.

They are based on my own experience, I hope they can be useful to you ! 🧙‍♂️

1. Separate explores in a specific file

Why : Your model file can quickly become huge. The ideal for your developers is to try to have small files responding to a use case. It’s also a good way to add governance to your code. Try to leave only the caching policy, the formats in your model file and of course all your includes for your different explores.

How : Create generic file with explore extension

The result :

In this file, you can add all the useful views for your explore.

Don’t forget, you need to include your explore path in your model.

2. Do not use the * wildcards

Why : If you have your explores in specific files, you can include all useful views for your explore.

If you specify all useful views, a new developer can easily navigate the file browser and easily know which files are useful for the proper functioning of the explorer.

How : Include the entire path, don’t include with the * wildcards

Tips : On the file browser, next to the view you want, you can retrieve the path by clicking “Copy File Path”

3. Develop a layering logic …

Aurélien Roblin

Analytics engineer - Indie Maker | NoCode, SideProject, Data science