Tableau is a well-established powerhouse in the business intelligence (BI) space. It has a ton of features, can produce advanced and highly customizable visualizations, and makes creating these visualizations easy for non-technical users. Although Tableau Server (self-hosted) and Tableau Online (Cloud) offer some online collaboration and web editing features, the true heart of Tableau is its desktop application. In fact, while many users at your organization can use the web app for analysis, every Tableau deployment requires at least one Desktop installation in order to build data sources, which isn’t possible on the web. The desktop application is extremely powerful and flexible for individual analyses, but isn’t focused on collaboration.

Looker is newer to the BI space, and it doesn’t challenge Tableau on its visualization chops. Just like Google Docs launched with a more limited feature set than MS Word, but made substantial improvements in the ability to collaborate in the word processing space, Looker made huge strides in the ability to collaborate in the BI space while focusing on a more essential feature set.

Collaboration is deeply integrated into every part of Looker. Looker is web-only, with no desktop application available — so there is no need to send files around or publish analyses to a server. All code written by developers is version-controlled with Git, meaning all changes to the underlying views, joins, dimensions, or measures can be peer-reviewed before going into production.

One of the standout features of Looker is how easy it is to share results. Every single change on a visualization, no matter how small (a color change, a change in the dimensions, sorting results, applying a filter, etc.), generates a unique URL that can easily be shared with anyone who has access to your Looker instance. Guaranteeing that other users are seeing exactly what you see is extremely powerful during development and when sharing results of any ad hoc analysis.

Because Tableau operates first and foremost as a desktop application, its collaboration options are generally more limited. Users can update data sources on the server, and those changes will trickle down to any workbooks using that data source. Analyses can be published for sharing, and can be edited by other users who have permission. But Tableau doesn’t have a built-in peer-review process before changes are put into production. It also doesn’t have a way to clearly understand what changes were made from version to version, or a way to share specific results from an ad hoc analysis without publishing a workbook.

You can get accurate results quickly with either tool, but Tableau is usually faster for exploration of totally new data. With Tableau’s drag-and-drop interface, you can start analyzing new data immediately, without much development work. This makes data visualization approachable to new audiences who aren’t proficient in SQL. But this also means users can gloss over basic data principles and create erroneous results with data they don’t thoroughly understand.

Looker puts data governance at the core of their product. Developers need to understand basic data principles and must write some code before any analyses can be run. This introduces some lead time when working with new data. But because everyone only has the option of working off of developed data sources, analysts and business users run far less risk of getting inconsistent answers.

Having at least one engaged developer is a critical part of any Looker deployment. At Brooklyn Data, we will often have multiple experienced team members working on a client’s Looker instance.

Tableau can produce a wide array of highly customizable visualizations. Users have fine-grained control over the look of their visualization, as well as tooltips and labels. But this flexibility comes at a cost. One of the most common challenges of Tableau is a steep learning curve. Even basic tasks like creating a bar graph can take some time to master. But with practice, you'll get the hang of it.

Looker has a more streamlined approach to creating graphics, which most users we work with report to be highly intuitive. Analysts can be less focused on the interface and more focused on the results. One of Looker’s strengths is that it shows results tables before building an analysis, and gives one-click access to the SQL query that Looker is passing to the database. This means an analyst can immediately see how Looker is interpreting the data they’re trying to visualize, and do a gut check on the data. Tableau hides detailed results sets behind a couple of clicks — still accessible, but harder to reach.

Because Looker so readily surfaces a results set, it is also much easier to build tables using Looker. Tables are notoriously challenging to create with Tableau, which tries to guide users towards more visual storytelling. In our experience, tables are rarely needed in a final dashboard, but they are invaluable during development and for ad hoc analyses.

If you have decentralized or constantly evolving data sources, Tableau is likely a better choice for you. Tableau offers more flexibility to easily pull in new datasets, and Looker’s more robust collaboration features will offer only limited value. If you have (or are moving towards) a centralized data warehouse, or have a larger team, Looker may offer you more value. 

Both can offer excellent value. Whichever you choose, Brooklyn Data has experienced analysts and developers to set your team up for success. Our ultimate goal is never to push one tool over another, but to recommend what’s uniquely best for you.

Want to chat more about what the best BI tool is for you? We’re here to help. Reach out: hello@brooklyndata.co.