Gartner has stated that the simplification of big data platforms is a primary objective for almost all analytics software vendors. This is largely to cater for what Gartner calls “citizen analysts,” the number of which is expected to grow at the rate of 400% faster than that of formally qualified data scientists.
For this reason, if you’re planning to launch an analytical solution and want to ensure it will be readily adopted, you’ll need to factor in a lot of design efforts for user interface (UI) and make sure it provides the best possible user experience (UX). Your end users should find the experience effortless, allowing them to interact with data in ways that they find intuitive.
To create such an experience, your UI design should follow certain key principles and best practices, some of which are common to any UX project while others relate purely to the presentation and visualization of big data insights.
The following guidelines fall into the latter category of big data visualization. They are observed and practiced by UX and big data consulting professionals and can be helpful if you are a stakeholder in buying or developing complex analytical software.
Define Your Users and Their Needs
If you are familiar with UX design, the need to research extensively into user requirements will not be lost on you. In the case of big data analytics, this research should be directed in such a way as to provide an understanding of what data (and in what quantities) users want or need to see.
Of special importance is the need to avoid overloading user dashboards with loads of data — a common mistake made when designers or developers fixate on demonstrating the richness of a solution’s analytic capabilities.
Your users know that there is an enormous quantity of data behind their tables, graphs, and charts — but the majority of them don’t want to see any more than is necessary to meet their objectives.
Design for Less Analytically Adept
Many enterprises will not have the budget to hire specialized analysts, and in any case, there are probably not enough skilled analysts to go around. This was highlighted by a recent Capgemini study into insight-driven enterprise, in which 39% of executives said a shortage of specialists is a major barrier to the use of big data.
With the need to make big data more accessible to the layperson then, the sweet spot to strive for in big data UX is one in which each user has an immediate view of the data that they need to monitor or interact with the most. This, of course, differs from user to user, so a winning UI will be one that can be customized to suit personal preferences — ideally by the users themselves.
Answer the Right Questions by Observing the Right People
In seeking UX insights through user research, some essential questions to answer include:
- Which should be the key metrics visualized to help users make decisions?
- How will different stakeholders (execs, managers, analysts) be using the data?
- Will the data be displayed primarily on large monitors, or on mobile devices?
- Do users need to monitor data in real-time?
- To what extent will users need to interact with the data presented by the application?
Regardless of how well you know your big data application users, try not to make assumptions about the answers to these questions. If you really must make assumptions, avoid the false-consensus effect by validating the assumptions with your target users before beginning UI development.
Neither software developers nor business analysts (unless they happen to be the people who will use your solution) are guaranteed to know what users’ real requirements are for a big data UI, and there is absolutely no substitute for time spent “observing” how users work, rather than simply asking for their views.
Use a Prioritization Structure for Dashboards
While you can gain a lot of valuable design guidance from observing and talking to users, you will inevitably make certain decisions based purely on design principles and a modicum of common sense. For example, once your users help you to understand what data is of most use to them, you will need to prioritize data presentation in some way.
In dashboard design, it’s a good idea to use tabs to separate different reporting elements, and to order those tabs in a structured way from left to right. The sequence of tabs should be structured to tell a story, which unfolds as the user navigates from tab to tab.
It also makes sense to prioritize the data displayed in each tab, with the most important metrics or results nearest to the top of the screen, preferably highlighted with color or a tile to help them stand out. This is one way to achieve the “sweet spot” mentioned earlier, enabling users to access their most important insights with little more than a mouse click or a single screen tap.
Don’t Create a Variety Show
It can be tempting to think that for non-technical users data analytics is a dry and dull topic and that, therefore, data displays need to be vibrant and flashy. However, going this way will only result in confusion, making data harder to interpret.
For instance, using a wide array of colors can seem like a good idea to enliven graphs and charts, but a single color gradient is actually much more effective in helping users to comprehend their data.
The same principle applies to selecting the types of visualizations to incorporate into a data analytics UI. A dashboard with a thoughtfully minimalistic layout, plenty of white space, and consistent visualizations will be more helpful and welcoming to users than the one with a garish mix of multicolored pie charts and 3D graphs.
Create “Levels” of Usability
Naturally, you will need to have some variety in your dashboards, because some chart types will not be suitable to display data in the way that’s most helpful to the users. The important thing is to put necessity, rather than creative ego, into the driving seat and strive for consistency as much as possible.
If people in different business disciplines, such as marketing/sales, leadership, and technical/analytical, will use your solution, you will possibly need to provide a more complex UI. If so, the best approach may be to design two or three levels of usability.
For example, non-technical users will probably fair best with straightforward, easy-to-digest dashboard visualizations. Management and leadership users are more likely to need access to menu-based reports and table selection. For analysts and data scientists, the third level of usability might include a combination of menus and keyboard commands, providing access to advanced features.
This Is How You Improve the Approachability of Big Data
Big data will always be complex, but that doesn’t mean user experience must be complex too. The guidelines provided in this article will help you consider how to make big data approachable for your targeted user base, which will be growing to include a greater number of people unskilled in analytical disciplines.
There is work to be done in many areas of big data processing technology in order to achieve this goal, but a great user experience is one of the critical prerequisites. Therefore, if your enterprise intends to build or buy a big data platform, it will pay off to keep UX considerations foremost in interface design — and to ensure they include what users actually need — both now and in the future.