Business Intelligence Best Practices - BI-BestPractices.com

Collaboration. Communication. Community.

 
 
 Printer-friendly
 E-mail to friend
  Comments
ADVERTISEMENT
Harnessing Visualization: Achieving Situational Awareness by Capturing and Sharing Thoughts

by Dr. Steve Roth
In avionics, flying blind means to fly at night or in fog using only instrumentation. When executives must rely mostly on metrics and dashboard-style data to pilot their businesses, they, too, are essentially flying blind.

Introduction

Hard data are useful and necessary as an indicator of organizational health, divisional performance, and trends. They’re also helpful for illuminating exceptions—instances in which something has gone off course and requires management attention.

However, the data in these applications represent endpoints: decisions that have already been made, money spent or earned, leads qualified, or sales clinched, for example. They show the whos, whats, whens, and wheres of a situation, but not the all-important hows and whys. As a result, they are less valuable for understanding a situation in-depth, plumbing the complexities behind the metrics, and deciding what to do next.

In a Harvard Business Review article, “The High Cost of Accurate Knowledge,” authors Kathleen Sutcliffe and Klaus Weber discuss the importance of people’s interpretations over facts and figures alone. In their research, Sutcliffe and Weber found that company investments in data accuracy yield a diminishing marginal return—while some degree of accuracy promotes organizational change, additional investment produces less, and eventually hampers change. They also found, contrary to conventional wisdom, that profit performance actually decreased as data accuracy increased. They note that in business, “information about the environment, though abundant, is seldom obvious in its implications.”

Data interpretation must come from people; the most valuable business intelligence in any organization resides in their heads. People can quickly overcome the Achilles’ heel of any BI system—missing, ambiguous, conflicting, or outdated data. They also provide “color commentary” that is vital to situational awareness but that cannot be summarized in charts and graphs.

Even so, it’s not surprising that many executives rely increasingly on data instead of people in order to try to gain situational understanding. People’s time is scarce; data, plentiful. Information systems have increased the available data exponentially, and advanced visualization tools make it easy to plumb them for important patterns. Large amounts of data are easier than ever before to collect, access, and use anytime from anywhere.

Visualization: A Medium for Capturing and Sharing Thoughts

Visualization makes it possible to capture, access, and share people’s thinking in the same way facts and data are collected and used. This gives executives and their teams a way to tap into this most valuable source of situational awareness. An online visual environment, combined with a flexible underlying information architecture, gives people a way to expose their thoughts and interpretations, along with hard data, to executives and team members.

Importantly, they do this without having to stop work, create a PowerPoint briefing, and present what they’re thinking. This implicit thought-sharing leads to a shared, real-time sense of situational awareness that exceeds by far the shared awareness provided by work-disrupting meetings and conference calls.

How can people read each other’s thoughts in a virtual environment? Consider an example from the physical world. Imagine stopping by your team member’s office to alert her to a budget discrepancy you noticed in the proposal you’re working on together. If you find that she is away from her desk, you can still infer a lot of information simply by viewing her workspace. For example, if a piece of paper next to her keyboard contains scribbled columns of figures related to the discrepancy you came to discuss, you can deduce that she knows about the problem and is also focused on it. A quick glance at her screen may show an open e-mail from accounting with a “budget problem fixed” subject line. From this shred of evidence you can hypothesize that the problem has been solved.

Even information that is not directly related to the problem can be helpful. For example, if you see that her coat and purse are in her office, you know she probably hasn’t left the building and can be paged if necessary.

All of this knowledge, which you learned rapidly from visual clues, gives you shared context. When you next see your team member, you can communicate much more efficiently. You can jump immediately to the how and whys: “So, how did accounting fix the budget problem?” Without these visual clues, you would have to start with a lengthy exchange of whos, whats, whens, and wheres: “I found a budget problem,” “I talked to accounting,” and so on.

In the same way people leave physical clues to their thinking and actions, people working in a visual, online environment leave traces of their thoughts and activities. Managers and team members can see and quickly understand these virtual clues without having to explicitly discuss all the details. I call these artifacts of someone’s thought process thought visualizations. In the right kind of visual environment, you can learn a lot about people’s thinking by looking at their thought visualizations: the data and other objects they work with and share, and their goals, tasks, interpretations, and thoughts.

One reason thought visualizations promote awareness more efficiently than do meetings and discussions is that meetings consist of sequential, spoken, written, and PowerPoint presentations. Writing reports, preparing presentations, and even verbally explaining whats, whens, and wheres interrupt work and require a great deal of effort to establish context. In addition, compared to visual communication, verbal communication is prone to misinterpretation, and is comparatively laborious to understand. Thus, communicating verbally is taxing cognitively for both the presenter and the audience. On the other hand, implicit visual communication is a natural consequence of people’s work, and gives a much more accurate view of what people are thinking.

Situational Awareness: A Whole Greater Than the Sum of Its Parts

Thought visualization is an effective way for executives to improve their situational awareness by understanding their people’s thought processes better. Leaders can also open access to their own thinking to more clearly share their priorities, questions, and intent. The exponential power of visual thought sharing, though, is achieved when people expose their thought visualizations laterally (to fellow team members) as well as up and down the line of authority.

This creates a common pool of thought upon which everyone can draw, as illustrated in Figure 1. People can view, share, and add to what their team members are working on, thinking about, and observing at will. As a result, the entire team moves from mostly individual understanding punctuated by periodic exchanges of words to melding the team’s thoughts into a continuous, shared understanding of the situation, including next steps. In many cases, team members interrupt collaborations preemptively, meaning they are aware of each other’s problem-solving activities and can inject insights that save time and correct errors.

alt
Figure 1. Vertical and lateral thought sharing

This common pool of thought is different from a “common operating picture” approach, which forces the team to accept a single interpretation. Shared situational awareness allows for disagreements among team members and their explanations of events. For example, team members may look at the same data showing a sudden drop-off in a region’s sales and interpret it in different ways. One person may think it is due to poor economic conditions that quarter; another may attribute the drop to a competitor’s new ad blitz; a third person may see it as the first sign that that entire product category is becoming obsolete because of a new technology being test-marketed in that region. Getting at the right interpretation, or the right combination of interpretations, is important in order to decide on the course of action most likely to correct the sales decline.

However, without visualization, these differing interpretations are often lost in the verbal fog of meetings before they can be explored in depth. Visualization captures and codifies differences of opinion so they can be analyzed.

A visual environment that makes it possible to pool thought in this way can take many forms, depending on industry, goals, and activities. However, it must have two characteristics:

  • Users should not have to interrupt their work to create presentations of their thoughts. Visualizations that show their thinking should occur spontaneously occur as a byproduct of their exploration and analysis.
  • Information (such as thoughts, assumptions, plans, questions, and observations) should be stored, viewed, worked with, and shared in the same form as hard, factual data. Because most team efforts involve both factual data and human insights and interpretations of the data, the two types of information must be associated—they should not exist separately.

Visualization-supported situational awareness is most useful in large organizations where teams of people collectively solve problems under pressure—e.g., teams at a multinational company responsible for introducing a new product, or teams of doctors at large hospitals.

Example 1: Plumbing the Hows and Whys of Sudden Sales Fluctuations
In the sales example, a global car company executive examining regional sales illustrates how this type of shared thought visualization environment works. In this situation, executives must track sales fluctuations and understand the hows and whys behind them. They must coordinate rapid corrective action by regional sales teams spread over different disciplines and miles of territory. Typically, this process is stovepiped. The executive might look at dashboard data, hold meetings or conference calls with the sales team, review reports from financial analysts, or review competitive intelligence. There has been no effective way to quickly combine all this disparate data and conjecture into a moving picture that explains the situation.

Figure 2 shows an example of a visual, shared situational awareness workspace that addresses this problem.

alt
Click here to open full-size image in a new window (large file!).
Figure 2. Thought-sharing workspace

The workspace is made up of frames—window-like containers that give appearance and behavior to data. Frames can contain hard data visualizations such as scheduling and resource charts, graphs, and tables; or visualizations of executives’ thoughts (such as sticky notes, questions, and comments).

The left side of the desktop illustrates the frames an executive is working on that she has made visible to her team members. In this case, the executive has zoomed in on the region experiencing the sudden sales drop; she is assessing why the drop may be occurring. Her desktop contains a message box in which she’s noted her critical questions (the main issues at hand): how the sales drop relates to marketing activities, competitive pressures, and so forth, and how to turn the situation around.

Her desktop also contains a map frame of the region, visualizing hard-data objects: in this case, her company’s dealers (rectangles) and competitors’ dealers (small triangles). The size of each shape corresponds to each dealer’s sales, so the executive can quickly see under-performing outlets and compare them to competitive activity in the immediate vicinity. Each of the symbols on the map has sales figures, anecdotal reports, observations from the sales team, and other data associated with it, which the executive can pull directly into a table or bar chart to visualize the details. This is important: each element in individual thought visualizations must be automatically linked to the underlying data, so users can drill into or roll up elements of the displayed map to rapidly retrieve the information they need.

The map frame also visualizes unquantifiable data: assertions (diamonds) are data objects she has created reflecting her interpretations and questions about the sales problem. Collectively, the map and its symbols reflect the executive’s understanding of the situation and her evolving plans. When shared, these convey her thinking about the matter at hand.

To the right of the executive’s primary work area is a set of 10 to 20 frames she has arranged from other users’ shared workspaces. These frames provide visibility into key team members’ data and activities. By tabbing into each shared frame to expose the data, an executive can quickly gain “topsight” on the thinking of her sales team. As a result, phone calls or meetings to discuss the situation are much more focused and efficient, since the team members are already familiar with each other’s ongoing work and areas of concern.

When a user (either executive or manager) sees important information in a team member’s workspace, he can copy it into his own, all while retaining the link to the underlying database. Information the creator updates will automatically be updated everywhere it has been copied. In this example, the regional sales director has proposed additional marketing dollars to remedy the sales situation. To support his proposal, he has shown current ad and promotional campaigns’ reach via dotted lines on his map workspace (to the right). The executive dragged and dropped the dotted lines into her workspace in order to visualize how the various ads and promotions that make up the marketing campaign relate to the sales activity at company and competitors’ dealers.

Collectively, the workspaces of the executive and her team comprise shared situational awareness of the problem that forms the basis for rapid understanding and informed decision making. A detailed history of the problem and its resolution is automatically captured for future reference—with no post-mortem report writing required.

Example 2: Urgent, Life-or-Death Decision Making
Another example of thought visualization in action can be found in the medical community, encompassing hospitals, research centers, and even pharmaceutical and insurance companies. In all of these organizations, teams of people with different areas of expertise must together make fast, sometimes life-or-death decisions based on large amounts of often uncertain and conflicting data and opinions. The 24-hours-a-day, 365-days-a-year nature of the medical field makes it very difficult to schedule time for teams to consult collectively about matters of urgency, such as a patient’s prognosis and treatment or whether to pull a drug from the market.

For example, when a patient at a large breast-cancer center has an abnormal biopsy, doctors (including her primary care physician, an oncologist, a pharmacist, and a pathologist) must rapidly agree on her prognosis and treatment. However, because of busy schedules and long, nontraditional hours, it may take several weeks to align their schedules. As a result, the patient must endure a stressful wait for her results and treatment plan.

In such cases, doctors can use thought visualization to quickly establish shared context or situational awareness. Thought visualization helps doctors communicate their data, assumptions, thoughts, and plans without having to schedule mutually convenient times to meet in person or by telephone. Figure 3 shows a visualization of a patient’s biopsy results and medical history.

alt
Click here to open full-size image in a new window (large file!).
Figure 3. Patient-at-a-glance

Because the pertinent data is presented in a quickly read visual, rather than the many pages of closely typed history that comprise a typical medical file, the doctors involved in treatment decisions can quickly acquaint themselves with the issues and immediately understand the chronology of her treatment.

Participating physicians view the visualization via computer. They note their observations, assumptions, experiences, thoughts, and questions via text boxes, virtual “sticky notes,” or scribbles. They can also attach pertinent research studies or findings as well as histories of similar patients. In this way, the physician creates a thought visualization that describes her assessment of the patient’s situation and her treatment recommendations.


In much the same way as the executive in the first example, the doctor in charge can tab through thought visualizations from other doctors on the team to rapidly understand what they’re thinking. Likewise, she can easily copy information from another doctor’s visualization into her own, while retaining the data’s link to the database. If she copies information from the oncologist’s visualization that he later updates, the information will be updated on her visualization as well. Any necessary phone calls or meetings to discuss treatment plans happen much more quickly, because the doctors involved already understand each other’s thinking. This means the patient receives a more timely diagnosis, and can pursue treatment earlier.

Elements of Shared Situational Awareness

Shared situational awareness depends on visualizations of the data and thoughts that are central to the matter at hand as well as adjunct thinking and information that could turn out to be meaningful. We offer examples of elements that feed shared situational awareness.

Explicitly Shared Objects and Events
In our first example, competitive intelligence (small triangles) and their attributes are often the focus of collaboration. They include types of objects and events such as potential customer populations where the competitor’s sales of cars featuring the new technology have already spiked, dealer receptiveness and positioning of the technology, reports of technical problems, media coverage, reports of a competitor’s supply chain problems (such as excess cars on dealer lots), news of other competitors’ reactions, and so on. Business analysts and salespeople provide most of these intelligence nuggets, which are then distributed to each user.

In our second example, collaborative objects and events are more wide-ranging, and include sticky notes containing questions about the patient’s openness to mastectomy versus lumpectomy, assertions with linked research findings on a promising new regimen for people in the patient’s age range and general health, and expert assessments on whether the cancer is likely to have spread.

In both cases, data appears automatically on maps, tables, and charts; users exchange data by sharing the visualizations that contain them. Others can then copy individual data elements to their own visualizations and thus contribute to analysis. Throughout this process, the individual data representations retain links to the database. Communication is expedited because team members share common reference points. However, because there can be a significant wave of data in a short time period (for example, at a major industry trade show, or during intensive patient treatment), filtering and interpretation can be difficult.

Level of Attention on the Objects
This is especially pertinent in the first example, where a large volume of data is coming in from potentially hundreds of sources. Each triangle representing competitive intelligence conveys whether it has been explored by a user (either by mousing over or by selecting it to reveal its attributes). A competitive-intelligence object has a yellow dot on it until it has been reviewed. At a glance, a team member can determine which competitive data the other members have not examined—whether due to information overload or prioritization. Knowing what a fellow team member doesn’t know presents opportunities for others to help out by exploring unexamined data and by calling attention to it if it is important.

Goals for Analyzing Objects and Events
The analysis process occurs primarily to answer key questions that serve the team’s goals and enable its tasks. Team members outline their goals in critical question frames. A leader’s critical questions expose what is important to him, why he is attending to some data over others, and which additional data would be valuable to share with him. Further, data can be associated with the goal boxes through drag and drop, so it is possible to see all the data that has been considered relevant to important questions. Thus, both data and thoughts such as questions and goals are made explicit and become part of the shared situational awareness.

Interpretations and Thoughts
To capture their conclusions about ambiguous, inconsistent, or incomplete data, team members add assertions to the map or chart and associate intelligence or research data with them via drag and drop. They also add summaries of conclusions and sketch paths indicating predicted market penetration, growth of disease, or other movements. This is another example in which shared situational awareness encompasses hard data and people’s thinking and interpretations. This often provides a filtering function in that team members can rely on each other’s interpretations of data to avoid attending to less important information.

History of Objects
Each data object created by a team member has a creation, modification, and copy history. Therefore, on any frame, it is possible to trace the actions being planned, the interpretations motivating them, the data on which interpretations are based, and the goals they all serve. It is possible to determine for any object which users first received it, annotated it, and transferred it to their own frames. Disagreements can be acknowledged and understood and corrections efficiently propagated because shared situational awareness includes the evolution and migration of information throughout the team. Histories can be the basis for collaboration—they reveal who has contributed to problem solving.

Ramifications of Shared Situational Awareness

The building of remote, shared situational awareness via the virtual capture and sharing of hard data and people’s thoughts leads to a change in the way people work.

  • Communication becomes mostly implicit and built into people’s work; it doesn’t require additional effort. In this way, all users become nodes that contribute to situational awareness. Without shared awareness, people must disrupt the work of others to ask questions. This often results in a cascade of interruptions as the first person refers the questioner to another person, and so on. With shared awareness, people dip into others’ data and thoughts at will, without having to disturb them. Consequently, information flows at an enormous rate throughout the organization.
  • In the absence of implicit shared awareness, leaders often rely on meetings, conference calls, and PowerPoint briefings to understand the latest organizational or patient developments. This reporting process diverts time and resources from critical tasks; and the information is often outdated by the time it is presented. Using thought visualization, users construct briefings on the fly by using a combination of their analytical tools, data, and individual context. This changes the briefing process from being a highly planned event (requiring hours of preparation and expensive, disruptive travel) to an ad hoc process that can be constructed in minutes and coordinated remotely.

  • Visual communication is efficient and much less error-prone than the primarily verbal communication of the past; people spend much less time gathering data and more time thinking about it and acting on it. Data, actions, and thoughts all are searchable, and captured for future reference.
  • In research exercises, visual shared situational awareness led to a six-fold increase in the percentage of time decision makers spent analyzing data (rather than gathering it); a 300 percent increase in the amount of critical information conveyed throughout the team; and a 300 percent productivity jump, since changes in the decision-making process meant users could act on information much more quickly.

Implementation Considerations

All new applications bring with them the need for some organizational tuning of both technology and business processes.

The visual thought-sharing workspace discussed here integrates easily with any hardware, software, and data streams. Users pull spreadsheets, e-mails, documents, and Web pages from existing applications. The primary adjustment is procedural: the more people use the application, the more benefits it provides.

People are usually slow to adopt applications that require additional work—for example, business intelligence applications that require them to write and upload summaries of their thoughts and activities. They tend to quickly adopt applications that make work easier, faster, or less tedious. As described earlier, a key feature of visual thought-sharing applications is the visibility of users’ thoughts that occurs naturally, as a byproduct of their existing work. Consequently, teams using thought-sharing workspaces are able to expose their thoughts, activities, and plans to others without having to stop and present what they’re doing. This is a powerful incentive to use the application at a time when the phrase “death by PowerPoint” has become common in the workplace.

In research exercises, visual shared situational awareness led to a six-fold increase in the percentage of time decision makers spent analyzing data (rather than gathering it).

Privacy and security are managed via permissions; people choose what they will make available to whom. There are two incentives for people to make more available rather than less. First, when users provide more access to their actions and thinking, the whole team has more shared situational awareness. This means individual users spend much less time presenting monotonous summaries of their activities and more time problem-solving. Second, the workspace becomes a moving picture of a user’s work; as such, it offers an opportunity for virtual face time with executives.

Executives also benefit when they share select views of their thinking with their teams. When team members know from day to day what is most important to company leaders, they’re likely to prioritize their work accordingly and quickly communicate related information that otherwise might not filter to the top of the organization.

Conclusion

While the idea of thought reading may seem futuristic, existing visualization techniques make it possible. In recent years, visualization has helped executives and their teams to easily see, analyze, and share large amounts of complex data. Now, visualization, built on a flexible information architecture, offers the ability to dip into people’s thinking, anytime, from anywhere, in the same way we’ve become accustomed to dipping into hard data.

The combination of data and thought visualization yields a level of situational awareness that cannot be achieved with the standard mix of dashboard data, conference calls, and meetings—and it does so as a natural part of people’s work. The result is tighter, more effortless coordination, increased productivity, and ultimately better management of problems and opportunities.

Dr. Steve Roth -

For more than 15 years Dr. Steve Roth has served on the faculty at Carnegie Mellon University's School of Computer Science. At CMU, Steve directed the SAGE Group, which performed groundbreaking research on visualization and intelligent user interfaces. The group designed an expert system that can automatically design visualizations and user interfaces based on the characteristics of the data, user sketches, graphical examples, and user preferences. Steve is also CEO of MAYA Viz, a company that creates software that helps organizations eliminate boundaries. He has focused much of his career developing innovative approaches to developing computer systems with which people can use and create information. Steve holds a Ph.D. in cognitive psychology and has published more than 50 papers related to visualization and user interface technology.