Business analysts need to evaluate and select the best data visualization tool to communicate key data findings to decision-makers with efficient, highly visual storytelling techniques.
The most common data visualization tools include Tableau, Power BI, Excel, Qlik, IBM Cognos and Sisense. Each of these tools can be useful as an organization matures its data and analytics practices to support its business needs.
"There is no single use case for data visualization tools in an organization and, in fact, most organizations have more than one depending on the end goal for each use case," said Kathy Rudy, chief data and analytics officer at ISG, a technology research and advisory firm.
How creativity can undermine clarity
It is essential to keep in mind the end goal of clearly communicating data insights to decision-makers when evaluating data visualization tools for a project.
"The crux of the matter is that clarity, color and chart selection with the art of graphic design help in the process of communicating complex information more accurately and effectively,” said Terri Sage, CTO at 1010data, a provider of analytical intelligence to the financial, retail and consumer markets.
The process of transforming data sets into visualizations -- scatterplots, pie charts, bar graphs, heatmaps, use of color and overall clarity -- is used to communicate the data findings. Teams also need to consider principles of human perception and cognition in visual design and learn how to tell data stories with the visualizations.
When creating data visualizations, input should be gathered from a UX/UI designer and the end user. Even if analysts may not be great at working with colors and fonts or selecting the optimal graphic type for depicting the message, they know what information needs to be conveyed.
"Most analysts do know what story they want the data to tell, which KPIs are important, and what looks good, even if they cannot create it themselves," Rudy said.
Start with the context
It can be helpful to assess different tools by starting with the context of what they do best.
Microsoft Excel is the go-to standard for quick analysis and charts to deliver quick value. It helps produce charts that are static with predictable outcomes. Rudy's team has built chart galleries to pull data input that is typically different each time. Excel also allows users to change the look and feel of fonts, axis definition, scale and color.
Tableau and Power BI are good for building dashboards based on user requirements that are fully defined and pulled from data sources such as CRM and ERP tools. These visualization tools can deliver rich information in preformatted visuals with options to drill into the charts for supporting detail. They also provide the ability to filter data on preset criteria, such as industry, region, customer type and product, to provide visual insights into the connected data source versus tabular reporting.
Tools such as Sisense can provide advanced functionality to build analysis and charting based on multiple data sets. Sets combine to create new ways of analyzing data and delivering insights, Rudy said.
It is also worth evaluating the idea of custom data visualization tools when extensive data prep is required. For example, if the data requires a calculation engine to curate and deliver insights, it might be best to create via the custom route. In this case, teams would take advantage of various data processing tools like python or R Tools for Visual Studio to create their own data analytics workflow.
This programming-based approach requires more development team resources but can also help organizations implement visualizations for less common scenarios that existing visualization tools may not cover.
Criteria for evaluating data visualization tools
Top considerations for choosing visualization tools include:
- ease of use and ability to map data to the visualization;
- type and quality of visualization output;
- availability of tool to run on laptop, tablet or server;
- cost considerations of the visualization tool and training if available;
- data import options;
- how the tool is installed and the system requirements for installation;
- interactivity of visualizations; and
- security of the tool relative to how the data is accessed.
Source: Terri Sage, CTO at 1010data, a provider of analytical intelligence to the financial, retail and consumer markets.
Size of data and data source to be visualized
It's helpful to start with a holistic view of data visualization needs when evaluating data visualization tools, said Priya Iragavarapu, vice president of global management consultant AArete's Center of Data Excellence.
"The size of data and the data source to be visualized is an extremely important consideration in which tool to be selected and if various tools need to be stitched together for effectiveness," she said.
For example, if the data is in cold storage, such as Amazon S3 cloud object storage, the performance may suffer even if Tableau provides that connector.
"Tableau is a great visualization tool, but if one places the onus of querying on Tableau, the performance and latencies get affected," she said.
In such a case, Qlik is a better tool since it has an built-in query engine to efficiently run a query on a large data set and/or cold storage, Iragavarapu said. Tableau is a good tool, but teams need to be aware of the strengths and weaknesses of the tools to get the best results, she said.
Another important consideration is the technology stack of the organization. If an enterprise is heavily vested in Azure Cloud or the IBM ecosystem, teams may want to start with Power BI or IBM Cognos, respectively, to simplify workflows. On the other hand, if the company does not have such a unified strategy, then they can mix and match tools.
Data pre-processing before visualization
Teams also need to consider the extent of preprocessing required before data is visualized. Ideally, visualization queries should directly query data and be able to filter, sort and aggregate data within the tool. This is not always practical, however, when complicated pre-processing is required, Iragavarapu said.
Each visualization tool includes associated preprocessing tools. For example, Tableau uses Tableau Prep for users to do their data prep work before data can be visualized.
It is also essential to consider the extent to which data will need to be prepared before users interact with the data and the extent to which users may need to prepare it for visualization. These two factors affect the performance of the visualization and the pace at which data is readily available for visualization.