Jeff business intelligence tool transition: from Metabase to Tableau
Jeff, the super service app based on Laundry, Beauty, Fit and Relax, started hardly working on data since the beginning.
First visualization tool was Metabase and, in early 2020, Tableau was added as a powerful tool to take the analysis to a higher level. Instead of choosing between Metabase and Tableau, we’ve discovered that both can coexist and make our data projects.
Jeff Data first steps
Metabase is the easy, open source way for everyone in a company to ask questions and learn from data, with a fantastic community of developers using the tool and whose versioning is constantly being updated. It was chosen as the business intelligence tool for the company due to their no-cost benefits (daily reports, easy dashboard creation and good visualization quality).
The most useful characteristics in Metabase are:
- Server data connections: PostgreSQL, MongoDB, MySQL and more. It’s possible to have several connections to different databases but only querying through one database at the same time
- Customize mode: allow to ask a more detailed custom question using the notebook editor, very useful for people with no SQL knowledge. However, it’s possible to write a new SQL query for harder questions
- Pulses: regularly scheduled reports to track most important metrics
- Easy dashboard creation: Metabase allows the users to create their own dashboards, adding filters and making them responsive.
- Metabase integration: integration of some dashboards into the most useful tool like backoffice
Metabase’s been the most used tool in the company due to all departments have employees checking data results. However, due to this huge number of employees running queries at the same time, we started having query running time problems due to queries run through databases in real time and it produced overloads while some users tried to run heavy queries at the same time. In addition, high complexity queries needed to be created by using SQL and running through one database at the same time was a problem.
These were some reasons why we started to think about new visualization tools, looking which one could fit better for us but not forgetting Metabase due to its potential. After a search through some interesting tools in the market, we decided to start with Tableau. In parallel with Tableau, our Data Engineering Team has develop a data warehouse to increase our query performance.
Adding Tableau as a tool
Tableau Software is an american interactive visualization software company. Tableau products query relational databases, online analytical processing cubes, cloud databases and spreadsheets to generate graph-type. Jeff analysts work on their own Tableau Desktop where each of them work separately. Once a project is ready, Tableau Desktop allow us to connect our Tableau Desktop to Tableau Server. Tableau Server is the place where different Jeff analyst can collaborate in a safety mode and share the dashboard to other Jeff workers. In addition, Tableau works with Tableau Online, a version of Tableau server hosted in the cloud. In this way, everybody could access to data with no previous installations.
The most useful characteristics in Tableau are:
- Huge variety of data connections: files (Microsoft Excel, JSON, csv,..), server (Oracle, Amazon Redshift, MySQL,…) and more
- Real time data and data extractions: working through extractions allow lower time execution. Data refresh can be scheduled once per different time intervals
- Tableau integration: integration our dashboards into existing applications
- Data blending: method for combining data from multiple sources. It brings additional information from a secondary data source and displays it with data from the primary data source directly in the view
- Calculated fields: allow the user to create new data from data that already exists in the data source. This new field is saved to the data source in Tableau and can be used to create more robust visualizations
- Drag and drop: despite of not being an easy learning tool, Tableau drag and drop allow the user to create visualizations easily
The capability to create data extractions solved completely our problem of running data in real time, reducing our query time execution and increasing users happiness. Tableau data blending solved data sources problem when using more than one at the same time and creating calculated fields improve our efficiency because no SQL was needed to create new KPIs.
Coexistence visualization tools
According to pros and cons of Metabase and Tableau, Jeff Data team decided to keep both tools to provide the best data service to the users. For these reasons, we still have several dashboards in Metabase while creating more complex dashboards in Tableau. Our final idea will be to migrate most important dashboards from Metabase to Tableau and setting Metabase for quick data checks, pulses reporting and for less used dashboards.
Teaching visualization tools
One of Jeff philosophies is that everybody can have access to the Data. Due to that philosophy, Jeff data team has been concerned about people using data through Tableau or Metabase so some training courses were prepared to the managers of each department to teach them the basic learnings to use both tools by their own.
Hope you all have enjoy this article!
Aitor Miota
Jeff Data Analyst
