Business Intelligence The Next Wave 3 Key Aspects of Useful Data Visualization Design
The only useful data visualization is one that can provide answers to important business questions. Data visualization is as much a science as it is an art. Not only does your chart, report, or dashboard need to have visual appeal, it should also help you make important business decisions. The only useful data visualization is one that can provide answers to important business questions. Any kind of data analysis should answer a key question that can help you make an important decision about the direction of your business. If you re creating a report for someone, you probably already know the question they want to answer, but do you know the important decisions that the answer will influence? You ve got to know your audience. You ve got to know the business. You ve got to know the elements of the choices that decision-makers face. Bearing that in mind, there are really three key aspects of useful data visualization design that will always help you facilitate better decision making. #1 ADAPT YOUR DESIGN TO THE WAY YOUR DECISION-MAKER WORKS Keep it simple. You may have a big monitor, or even multiple monitors, and there will be a temptation to fill an entire screen with charts and graphs like ornaments on a tree. This is not the way your end user works best. Keep in mind that they re on a laptop screen, or a tablet, or maybe even a smart phone. As a rule of thumb, if you ve got a 20-inch monitor, cover up half of it, and design your dashboard to fit within that space. It s better to create multiple pages or tabs than to have an overly complicated layout. Ask about visualizations they re already using to make your design familiar to them. If they use maps, use maps to answer their questions. Do they use ternary charts? Do they use schematics? Do they use flow maps? Don t use default colors. Use the company colors, or colors that are industry standard. This is a small, simple thing you can do that makes a subtle but distinct impression of quality and thoughtfulness.
#2 PROVIDE INTERACTIVITY AND A WAY TO TEST ALTERNATE THEORIES When you re designing your data visualization, don t think of them as static or dynamic reports. Data visualization dashboards should be control panels that can be touched and manipulated by the end user. They re decision sites with drill downs that can tell a full story or guide you through a series of progressively more specific questions. This is where a more modern data visualization tool like Tableau, TIBCO Spotfire, or Microsoft PowerBI can come in handy over built-in reports from off-the-shelf software tools or what you can get out of Microsoft Excel spreadsheets. In addition to answering the explicit questions they want to know, interactivity gives your decision-makers the opportunity to pose what-if questions and create scenarios for themselves with the visualizations you provide. Ask them what constraints they may have, and then allow them to filter down to those constraints and let the visualizations show them what would happen as they tweak variables. This lets them choose from multiple options, or plan for many different scenarios. This is where you start to see something called self-service analytics really come to life. #3 AGILE DEPLOYMENT Agile software development practices help a great deal in data visualization design, because you can provide regular delivery in short spurts, continuous improvement, and reprioritizing the things you need to create or tweak. Instead of weeks and months between revisions, you can collect feedback and collaborate to make more rapid changes to your visualizations without creating a disruption in the decision-making process.
OTHER THINGS TO KEEP IN MIND The people who know your business best and have to make important decisions often still face a challenge with slow, expensive, and inflexible reporting systems that provide inadequate answers to the wrong questions. By contrast, the promise of today's modern visualization and reporting tools is to improve decision-making processes by presenting key data in an interactive, easy-to-understand format that allows the decision maker to rapidly test possible decisions with predictions. If you haven t done so already, I would highly recommend licensing a more advanced tool. Show your end user the bad data, the missing data, and any uncertainties that are coming from the data. Don t hide it! The important thing here isn t to create doubt, but to reassure and to set an expectation that there will be distortions and imperfections, but that they will not impact the ability to make a decision. Stephen Rasey BI Architect stephen.rasey@entranceconsulting.com Dr. Stephen Rasey is an international petroleum geoscientist and economist with 34 years in exploration prospect generation, risk and volumetric uncertainty analysis, economic evaluation, portfolio management, and software development in the petroleum business.
Office 365 + SharePoint Custom App Development Data Management & Analytics Application Management entranceconsulting.com 888.343.5669