Admin overview
An overview and list of resources for Metabase admins.
The good news is that all you really need to do to get started as an Admin is connect Metabase to your data sources, invite people to your Metabase, and ask them to build some dashboards. But you probably have bigger plans, so we put together a list of resources that we’ve written to help you solve common problems that orgs face when standing up an analytics platform.
This article assumes you’re either on a Metabase Cloud plan, or that you already have your self-hosted Metabase up and running.
Basic setup
From the main Metabase navigation, if you click on the Gear icon > Admin Settings > Settings page, Metabase lists some tasks for you to complete:
Get connected
Curate your data
Again, the most important things are to add your databases (or get someone else who has your database connection details to add them), invite people to Metabase, and ask them to build some dashboards. Eventually you’ll probably want to get reports emailed out and whatnot, but the main thing is to just get everyone onboard.
Set up data and collection permissions
Though Metabase simplifies permissions, managing permissions is just inherently complicated. In addition to our docs on permissions, we wrote an article on managing people in Metabase, and a whole guide to permissions:
Encourage people to use the query builder
While your analysts will probably gravitate toward SQL queries, it’s generally better to use the query builder. The reason is twofold:
- With questions built using the query builder, people can drill through the data.
- Query builder questions are also easier for people to copy and play around with. No one needs to know SQL (though query builder questions make a great stepping-stone toward learning SQL).
Alternatively, you can use a SQL query to build a model, and let everyone use the query builder to slice and dice the data in that model.
Annotate data with events
You can add events to collections to capture institutional knowledge: when you shipped important product updates, had a sale, ran an ad campaign, and so on. Events will show up on charts, so people will stop having to ask (again) why the data spiked back in April.
Keep things organized
Once people figure out what they can do with Metabase, the questions and dashboards will start piling up. You don’t need to (and probably can’t) tame the chaos completely, but that’s okay. We wrote up some strategies for fighting entropy as your Metabase grows.
Learn the different ways you can share data
You’re probably going to have to share charts and dashboards, both within and outside of your org.
Embed charts and dashboards
If you’re using Metabase to embed charts and dashboards in your app, that’s a whole other game, and we have some articles for you:
- Strategies for delivering customer-facing analytics
- Publishing data visualizations to the web
- Multi-tenant self-service analytics
Advanced administration
If you really want to get serious about running Metabase, we’ve got you covered:
- To preload a Metabase with questions, dashboards, collections and so on, check out: Serialization
- For running Metabase in production, check out Metabase at Scale
- Tips on how to speed up dashboards: Making dashboards faster
- To see how people are using your Metabase, check out How to keep tabs on your data
- To program Metabase, check out Working with the Metabase API
Tell your team about Metabase resources
There’s no need to develop a whole course here, but even a single session where you demo Metabase with your data can help your team make the most of their new tool.
We (the Metabase team) create a lot of resources to help level people up, so make sure your teams know about them.
- Tutorials and guides: Learn Metabase (you’re staring at a Learn article right now; you are learning).
- Metabase reference: Metabase docs.
- Forum where people (like you) can ask and answer questions: discourse.metabase.com.
- Community posts, where data practitioners (also like you) share their stories and tips: Community Stories.
- Blog: Metabase Blog.
Give people a tour of your data
Because people have access to these resources, and can learn on their own, it’s more important for you to walk people through what data you have, and where people can find that data. You might not know about all of your data, so find out who does and ask them to educate the relevant teams. Some orgs host data happy hours, or lunch hours, or just recurring meetings where people present their analysis and show people where they can find the data they used. Or better: have people build a model in Metabase, and present on the model they made, and the kinds of questions that model can answer.
These meetings can be fun (for some definition of fun) and surprisingly effective at circulating institutional knowledge. The main things to emphasize are:
- How to take existing questions and find the underlying data.
- That it’s safe to play around with existing questions. Starting from existing questions is a great way for people to teach each other not only how Metabase’s query builder works, but what data your org has to offer.
- That they can X-ray tables to get a starting set of questions to play around with.
You can also create a dashboard in your Metabase and use text cards and charts to present your teams with an overview of your data.
Next: How to run Metabase in production
If you're self-hosting Metabase, here are some benchmarks and best practices.