Although Tableau can work with a wide array of data inputs and can create stunning visuals, it takes a thoughtful approach to how you manage your data to make the most of Tableau’s analytics engine.
In this blog, we will go over Tableau’s strengths and limitations, and then dive into key considerations for any business looking to use Tableau to take their organization’s analytics to the next level.
Tableau is a market-leading visual analytics platform and has been named a Leader in the Gartner Magic Quadrant for Analytics and Business Intelligence Platforms each year since 2012. Aided by proprietary features like Tableau’s VizQL analytics language and hyper in-memory data engine, Tableau is built to empower business users to get the most insight from their data. Designed with native visual best practices, Tableau allows users to seamlessly create powerful, interactive dashboards in an intuitive interface.
Relative to other business intelligence (BI) platforms, Tableau excels at creating sleek, crisp visualizations that can be built quickly by analysts and easily digested by end users. The platform particularly excels with geospatial analytics, allowing for users to derive insight from physical data points. The platform is designed to work with a wide variety of data sources, and connection options range from cloud and on-premises SQL databases to Excel spreadsheets, Salesforce, and even PDF files.
Tableau often works well in a minimally governed or a highly-managed data environment —it’s the murky middle zone that can cause problems for developers and business users alike.
If your organization has a specific use case that requires one to two tables of information that you want to bring to life with Tableau, those tables can likely be ingested and handled by Tableau’s data engine without much trouble. Similarly, if your organization already has a managed data warehouse with conformed facts and dimensions, your data can come to life in Tableau with minimal fuss—if anything, developers can create an extra layer in the data warehouse that queries only the data they need to bring into Tableau.
However, we’ve identified three scenarios where Tableau may not produce what you need:
In each scenario (broken down below), a common takeaway is to be mindful of the number and type of data sources in your dashboard.
Because Tableau’s data model is limited relative to other BI solutions, much of the data management legwork should be handled outside of the confines of Tableau’s software.
If you have a working data warehouse in your organization, Tableau can generally handle a large volume of data. The key to working with Tableau in conjunction with a data warehouse is to push as much transformation and business logic as possible into the data warehouse and away from Tableau itself. The technology your data warehouse is built on is not so important—Tableau works well with cloud warehouses like Snowflake as well as on-premises SQL servers—what’s important is to minimize the number of tables joined in Tableau when building your front-end application.
Tableau rolled out its relationships feature in its 2020.2 release, which makes it easier to include disparate datasets in your data source, but performance concerns remain if you are trying to relate more than a few medium-sized tables together in one data source.
To get around these performance limitations, you have a couple options:
Because minimizing joins is important in Tableau, we recommend working with a de-normalized ‘star schema’ data warehouse structure rather than the relatively more unwieldy entity/relationship structure. Although the approach of pre-joining tables in your data warehouse requires an additional layer within your data warehouse, the performance gains you realize from consolidating your tables in the warehouse rather than in Tableau should more than offset the cost, allowing you to truly maximize the value of your data.
You certainly do not need a full data warehouse to get beautiful and insightful analytics out of Tableau. Tableau has a few key features, such as the ability to read PDF reports and the Data Interpreter automated cleaning feature for Excel/PDF files, that make it easy to take a single report and bring it to life using Tableau’s analytics interface with minimal manual intervention. If you have one or two key reports that you receive from another business user or third party and you want to see it visualized in a more intuitive and insightful format than a typical Excel or PowerPoint report, Tableau is a strong choice.
The beauty of using Tableau in this scenario is that the platform makes it very simple to go from receiving a report to creating a dashboard: just load the data in, define a relationship to any other tables you’ll be using, and get to work.
A key consideration with this approach is to understand when your dashboard shifts from a ‘one-off’ style build, where you are using only one or two key reports in a dashboard, to more intensive data requirements. In these scenarios, look to other tools, such as Tableau Prep or a data management solution, to help manage your workload and perform any necessary data transformations. We explain more about this scenario below.
Tableau Prep lets you manage your workflow and is useful if you have more than a couple different data sources to bring together in your dashboard. While Prep automates the joining, field renaming, and pivoting you’d otherwise need to do in a database or in Excel, it is not a data warehouse replacement and is best used strategically rather than as a blanket solution to data management. Photo Credit: Tableau
If your organization lacks a formal de-normalized data warehouse but still requires several data sources or robust transformation logic to prepare your data (such as pivoting tables, renaming columns, joining tables, doing extensive field calculations), a more thoughtful approach is required to deploy Tableau effectively. In these scenarios, it is often helpful to employ a tool to help ease the transition from raw data to report-ready data sources.
Tableau Prep can help in some of these cases—and can also be effective with some scenarios where data is minimally governed as well. Tableau Prep is available with any Creator License (the same license that is required to use Tableau Desktop) and provides a workaround for Tableau’s relatively limited data modeling capabilities.
With Tableau Prep, a developer can take datasets and ready them for analysis in Tableau. All the pivoting, field renaming, and joins can be achieved with Tableau Prep’s drag and drop interface, resulting in a unified dataset that can be saved to Tableau’s proprietary .hyper file format, compressing large volumes of data into a manageable file size that Tableau can work with.
Although Tableau Prep can theoretically handle many disparate datasets, it is not a data warehouse replacement and is best served working with a small handful of tables that are not too large. Tableau Prep is often useful for automating by-hand work that an analyst does in Excel to prepare the data for the front-end application.
Sometimes a more formalized data management approach should be considered. These can run the gamut from a one-off Python or R script that automates data cleansing (although these should be deployed in a sustainable manner), using a data management tool like dbt to transform your data, or developing a formal data warehouse.
Some are intimidated by the up-front cost of these data management solutions, but over time, a formalized data management strategy will help you move out of the ‘murky middle’ of a high-data/low-management environment into a more sustainable data management approach.
Regardless of how your data is governed, there are some other key considerations for getting the most out of your Tableau deployment, each of which are inter-related.
There are several approaches to using Tableau. The more comfortable your organization is with managing data, the easier it will be to find a sustainable and functional solution to leverage Tableau’s analytics engine and visualization capabilities.