Spotlight on: data visualization tools

Data visualization tools are an important part of the data analytics stack. Once your ETL pipeline has been tested and works reliably, and the processed data is ready in your data warehouse, you need a way for your teams to access, understand and report on it.

In theory, data visualization software should present data in a way that’s easy to interpret, and therefore actionable. But not all data visualization tools are equal.

So how do you choose the right software for your team? Let’s take a quick look at the platforms on the market today.

Power BI

Microsoft Power BI is a cloud-based data visualization tool designed primarily to be used with business intelligence data (hence the “BI”). It can be used for reporting, self-service analysis, and predictive analytics. The idea is that Power BI functions as a centralized repository for all of your business data, and can be accessed by all the members of your business team. Through its user interface, you can create reports and share these with others in your company.

Power BI can interface with external data sources via its officially-supported integrations. These include MailChimp, Google Analytics, Salesforce, and other Microsoft products such as Exchange and SharePoint. When importing data, Power BI performs some cleaning so that the data set can be better ingested and understood.

Often perceived as the next level up from Excel, Power BI provides you with the ability to collect, examine, and visualize data from across your business, enabling you to make more informed decisions. If desired, it also offers predictive analysis, using ML to predict future outcomes. Power BI can handle extremely large quantities of data, allows users to create their own customized dashboards, and provides an interface that’s much easier to use than complex spreadsheets.

Tableau

Salesforce Tableau offers many different visualization formats, including a mapping feature which allows you to create color-coded geographical maps to show regional information. Tableau also supports hundreds of data formats for import, including Google Analytics, Google Ads, and CSV.

Since Tableau consists of a number of different versions, let’s explain the differentiation between these:

• Tableau Online – SaaS hosted on Tableau’s servers, enabling you to create and share data, reports, and dashboards across your company.
• Tableau Server – Can be hosted on local servers or AWS (Amazon Web Services).
• Tableau Desktop – A desktop app that connects to your data to carry out analytics, and can also be used to create reports, dashboards and storyboards.
• Tableau Public – Doesn’t include automation, so you can’t schedule reports. It’s also limited to local data sources, and to 10 GB of content which must be saved to the server rather than locally. Perhaps more importantly though, any workbooks are publicly accessible by anyone online and there are no security measures.
• Tableau Mobile – A free app for iPad, iPhone, and Android enabling you to create a dashboard once, then view or edit it on other devices.
• Vizable – A free app for the iPad enabling you to perform analytics and share the results via social media, email, or instant message.

Looker

Looker was acquired by Google in 2019 and is now part of the Google Cloud Platform. It has a unique approach in that it requires you to perform a data modeling step before generating any data visualizations. Moreover, Looker uses its own proprietary modeling language, LookML, so you need to invest some time in learning to use it.

LookML is essentially an abstraction layer on top of SQL, describing the dimensions, aggregates, calculations, and data relationships inside a database. Important concepts are ‘views’, ‘explores’, and ‘models’.

A view is a set of fields that are linked to a table, an explore is a view that can be queried and also defines the relationships to other views (called ‘joins’), and a model is a custom portal consisting of a set of related views and explores. LookML’s three-step modeling process starts by defining views, then combining them into an explore, and finally gathering explores into a model which can be passed to the BI team.

Looker works well if your data team appreciates its data modeling ethos, but its uniqueness could divide opinion within an organization depending on who needs to use it to access the data. The important thing is to understand how the system works, and whether your team is on board.