In the information age, data is often one of a company’s most valuable assets. The derived value is usually multi-faceted and realized in different ways throughout the organization. Technologies such as data warehousing, business intelligence, analytics, and machine learning are used to create value from billing and operations through to the development of new products. In these diverse facets, consumers of the data within the organization can include the line of business users, analysts, data scientists, executives, and even the customer.
Data visualization can play a vital role in all of these scenarios; it is where the rubber meets the road, or rather where the insights meet the eyeball. With so many diverse applications and users; however, how can you choose the right technologies and capabilities for realizing the potential of data visualization to reveal, explore, and explain your data?
Looking to the market for the answers can be overwhelming. There has been a proliferation of data visualization related technologies over the last decade: in addition to well-known offerings from players such as Tableau, PowerBI, and Spotfire, there are many other related systems that promise to address some part of the visualization challenges through offerings such as automated feature discovery, integrated analytic workflow, scalable data access and management, assisted chart creation and flexible dashboarding. All of these offerings are entwined across the surge of interest in data visualization from different perspectives, including data science, business analytics, enterprise reporting, data journalism, and infographics.
Viewing and understanding an effective data visualization is infinitely more comfortable than creating one
If data truly is one of your organization’s most valuable assets, you cannot afford to take a one size fits all approach to data visualization. By all means, use traditional vendors to handle the mainstream line of business needs such as enterprise reporting, but don’t settle for low-fidelity aggregated snapshots of your KPI stats and distributions in pretty charts and call it a day. Many enterprise data lakes contain precious insights that can only be uncovered through high-fidelity views that show the fine-grained patterns that get lost in larger aggregations. Additional insights can also be obtained through the exploration of correlations, clusters, trends, and outliers, which cannot be predetermined by any data warehousing and reporting team.
It is a challenging task to build up this capability. Viewing and understanding an effective data visualization is infinitely more comfortable than creating one. Creating data visualization systems on top of enterprise data is an inherently integrative discipline that depends on tight synchronization between many different capabilities. It requires expertise in related domains of business knowledge, data management, and access, full-stack application development, graphic design principles, user interface development, and UX. Data visualization itself is an often poorly-understood discipline with decades of research into layout algorithms, properties of human perception, color theory and design, information design, and more. To do it well requires much more than just the ability to program a bunch of charts.
In addition to being highly integrative, data visualization is also highly iterative. It requires a continuous cycle of finding and viewing aspects of your data, which in turn gives rise to incremental or revolutionary insights that drive the further investigation. Sprinkling all of the required capabilities across the organization makes it difficult for them to obtain a high degree of synchronization necessary for an integrated and iterative solution. A much more effective solution comes from bringing those capabilities tightly together within a single team to allow for a more coordinated approach.