Dashboarding and the concepts of real-time, easily accessible information for the purposes of decision making, are not new concepts. In fact, some might be surprised to learn that origins of digital dashboarding for decision support systems stem back as far as the 1970s. Even with this great heritage, it is still evident that for many organizations, the dashboard is a highly misunderstood, poorly constructed infographic that looks better as part of a software demo than a day-to-day management tool.<\/p>\n
A dashboard in its simplest form is intended to present a collection of complex and relevant data through a series of visuals. The most familiar dashboard is the automobile instrument cluster. The purpose of the instrument cluster is to give the driver updates on various conditions: speed, direction of travel, remaining fuel, time of day, vehicle mileage, etc. By glancing at this information display, the driver can make quick, effortless decisions to stop for fuel, increase or decrease speed, or consider upcoming preventative maintenance requirements and needs. Similarly, a business dashboard \u2014 especially one tuned for a technology environment \u2014 should allow information to be digested as quickly as possible so that effective decisions can be made.<\/p>\n
The key difference in an effective dashboard system is the ability to reduce the complexity of the information being presented, and allow the human mind to subconsciously take over and accurately compute \u2014 recognizing right, wrong, good, or bad conditions instantly so that appropriate action can be taken.<\/p>\n
Figure 1a depicts a series of books out of order. Assume these books are manuals, and part of a set of commonly used reference materials. If individuals simply take a book, and return it ad-hoc, a simple question such as \u201cWhich book is missing?\u201d, or \u201cWhere is the system diagram manual?\u201d can actually become a small challenge. Don\u2019t get me wrong, it\u2019s not taxing. However, it does require conscious effort and brain cycles to answer.<\/p>\n
In order to overcome situations like these, most people number or alphabetize a series of manuals in a set, as in Figure 1b. The system depicted in 1b seems simple, easy-to-use, and it reduces complexity. In many organizations, a numbered or alphabetized system is thought to be best-practice, and so it remains for an organization. Similar to most dashboards, this system can still be improved but most don\u2019t take the effort to do so because they already believe that further optimization is not necessary. A good dashboard should be optimized for ultimate simplicity. If we were to return to the images in Figure 1, Figure 1c would represent a successful book ordering method that reduced complexity. The simple technique of drawing a line across the spines of the book is simple, yet it allows the brain to take over subconsciously, allowing one to immediately understand which book is missing \u2014 if any \u2014 and where it should be located, in order. This is the true power of visual information.<\/p>\n
When working with organizational dashboards, understanding the purpose and starting design from the ground-up is critical. The business must take appropriate time to ensure that the questions being answered by the dashboard are relevant, and relate to a well thought out series of objectives. We start the process by asking questions based on KT\u2019s Decision Analysis methodology and listing responses:<\/p>\n
When responses have been generated, it is best to take a second look and arrange them in the categories or the natural groupings they represent. This way, when a discussion is complete, it is easier for everyone involved to step back and ensure that the objectives are appropriate and balanced for the business needs.<\/p>\n
Not everyone requires the same amount of visibility into an end process. The CIO of an organization, for instance, might be interested in how Service Level Agreements (SLAs) are being met, and what the current cost of service is, whereas an application administrator might be more interested in the free disk space percentages on each of their servers, and the current uptime of critical assets. Common pitfalls at this stage include trying to design a single catch-all metric; including too much information; and wishing information to be a part of the dashboard simply because \u201cwe have the data.\u201d The design of a dashboard should be focused around an individual, or a group of users that are empowered to make decisions and act upon the information that is being presented.<\/p>\n
The goal of Step 2 is to prioritize and connect the objectives gathered in Step 1 for each group or individual that has been identified. Ideally, this step should be done with the members of the group for whom you are designing the dashboard, so that they may identify gaps and be able to share their needs with the designer.<\/p>\n
This step should also include a rigorous clearing-out of unnecessary metrics. A recent CIO customer of KT inherited a metric pack of 192 indicators of business success. When the CIO asked the questions, \u201cWhat do I do if this goes up?\u201d and \u201cWhat do I do if this goes down?\u201d, those metrics were reduced to less than six.<\/p>\n
Designing and selecting appropriate metrics is a task that must be embarked on with great care. It is listed as Step 3 for good reason; too many organizations select metrics, and then build a process around them \u2013 leading to frustration and poor business results. Table 2, as shown below, details the questions required (based on KT\u2019s Decision Analysis methodology) to create effective metrics.<\/p>\n
The most important thing to remember in this step is that although a wide-variety of metrics may be available to measure best-in-class service, not all of them may be appropriate or tell the complete story of your organization\/department\/group\/product. Consider the value to both the customer and the business of a given metric.<\/p>\n
A common pitfall in metric design is viewing metrics in isolation. Viewing metrics in isolation can lead to incorrect conclusions and drastic decision making, as you are relying on only one metric to tell a story. Although some metrics have been researched and have been shown to be quite accurate in depicting a story, such as Net Promoter Score (NPS) or more recently its successor, the Customer Effort Score (CES), these have been the exception, and not the rule. The best practice in metric design is to layer a series of metrics, thus using multiple sources of inputs to create a picture to aid in decision making rather than relying on a single metric and making incorrect assumptions. Table 3, shown below, depicts a series of metrics: a particular condition one might observe; the common thinking most would jump to after viewing the metric in its condition; and a sample reality that shows just how poorly the metric told the complete story.<\/p>\n
Another common pitfall in metric design is not spending enough time in Step 2, that is getting to know your audience. By doing this, a designer might choose a metric that they don\u2019t realize is easy to \u201cgame.\u201d When a metric is \u201cgamed\u201d, inputs into the metric are purposefully manipulated to distort the picture, thus making an organization\/department\/group\/product seem better than it is. Table 4 is a list of examples of metrics that are well intentioned, but upon closer inspection, easy to \u201cgame\u201d.<\/p>\n
Metric design requires careful consideration and resolution of all challenges before proceeding to Step 4 and creating a dashboard. Issues to consider include:<\/p>\n
The next step in the dashboard creation process is to actually start to visually organize and display data. In order to do this well, it should be designed to tell a well-connected, and easy to follow story. The dashboard should contain graphs, charts, and visuals that have meaning when viewed in close proximity. The visual should allow for the least amount of conscious thinking required to understand and then take action on the data it represents.<\/p>\n
To ensure that charts are quickly actionable, it is helpful to perform some thinking in advance to determine triggers and control limits. Triggers are a threshold value that automatically begins a pre-defined action if a limit is reached, or allow anyone viewing the dashboard to manually make a decision and begin pre-determined next steps to resolve the condition. Control limits are minimum and maximum allowable values of a metric for it to be considered \u201cin-control.\u201d A value that is outside these pre-determined limits is therefore considered \u201cout-of-control\u201d and by making the condition visible it is easy to determine that action must be taken to resolve the condition.<\/p>\n
It is important when selecting a chart design to understand the level of detail required for effective decision making based on the intended audience. Figure 2 shows a series of different charts all depicting the same risk conditions. Consider the popular \u201ctraffic-light\u201d chart, depicting Red, Yellow, or Green, depending on the status of the metric it is linked to. This type of chart has two built-in pitfalls. Firstly, most will only take action on items that are in the red zone, incorrectly suggesting that non-red items are not critical and therefore do not require effective action. Secondly, the chart does not show trend information. In reality, the underlying data could predict that something today in yellow will turn red tomorrow. Care must be taken in selecting chart type so that a balance between visually pleasing and functional information can be displayed.<\/p>\n
A difficulty often seen when designers are completing Step 4 is that it contains so many graphs and visual displays that it begins to resemble something you would expect to see at NASA\u2019s mission control.<\/p>\n
Step 5, which I jokingly call \u201cDesigning for Digging,\u201d is about prioritizing the information developed in Step 4. The key here is to design a dashboard that is layered. The top layer could contain the most high-level information, and by drilling down, or selecting a group or chart, one could uncover the source of the data, related graphs, or other layered metrics to assist in performing effective actions.<\/p>\n
An anecdotal best-practice is that a dashboard layer should have no more than six (6) important pieces of data displayed at any one time; this stems from research on how many discrete channels of data the mind can keep track of at any one time. It may seem like a limited approach, but if designed for digging, a dashboard should provide the user with the ability to get to the most useful data quickly when required. An example of drilling is shown in figure 3, illustrating how additional information can quickly be digested if required. The drill down approach to dashboard design allows one to balance the abilities of layered metrics to tell a story, quickly aiding an under-pressure decision maker to control the situation effectively.<\/p>\n
Recognize the dashboard as a tool to help aid in the decision making process. This is simply modeling the business, removing distraction, and focusing on key data to facilitate accurate and timely decision making \u2014 which may need to be made in a time critical setting. Without taking a step back before developing a dashboard and recognizing the system for what it is, and what it is not, one could easily fall into the trap of thinking that quantitative data will always trump qualitative data. Good decision makers should always use all the tools available to them, not just be transfixed by a single tool, and expect it to do all the work for them. The allure of a pretty dashboard can trick people into forgetting that some of the data could still be erroneous, either due to collection error, \u201cgaming of the system, or an assumption in metric design being broken due to a changing environment. Whatever the case may be, the dashboard system and design, like other processes, must be regularly reviewed for accuracy, relevance, and alignment to business objectives.<\/p>\n
Happy designing!<\/p>\n
\u00a0<\/p>\n
Kepner-Tregoe is the leader in problem-solving. For over six decades, Kepner-Tregoe has helped thousands of organizations worldwide solve millions of problems through more effective root cause analysis and decision-making skills. Kepner-Tregoe partners with organizations to significantly reduce cost and improve operational performance through
\nproblem-solving training, technology and consulting services.<\/p>\n","protected":false},"excerpt":{"rendered":"
The Power of Visual Information Dashboarding and the concepts of real-time, easily accessible information for the purposes of decision making, are not new concepts. In fact, some might be surprised to learn that origins of digital dashboarding for decision support systems stem back as far as the 1970s. Even with this great heritage, it is […]<\/p>\n","protected":false},"author":8,"featured_media":39037,"template":"","white-paper-type":[],"class_list":["post-39036","white-paper","type-white-paper","status-publish","has-post-thumbnail","hentry"],"acf":[],"yoast_head":"\n