Designers must approach dashboard design with a strong emphasis on the user; however, deciding which elements impact on the user experience is the first step in producing a device, system, user interface, or any other product intended for human users. When designing dashboards, these elements can be thought of as a series of questions, which help focus the designer on the key issues facing them when producing displays that allow the users to carry out their intended activities both quickly and accurately. Some of these questions are shown in the brainstorming session; you will no doubt think of many more important questions to ask yourself and your design team that are based on general design needs and those specific to your current project.
Establishing a list of questions, whether this is in the process of brainstorming or any other idea-generating process, can be used to form a rudimentary checklist, which the designer can refer to over the course of a dashboard design project. In every project, the designer must contend with a number of different factors, variables, people and needs, collectively they can draw attention away from some of the key considerations and the designer loses focus on what really lies at the heart of the intended user experience. A checklist or set of design questions provides a simple and quick reference to ensure the project never veers too far away from the original, key considerations.
Overview of the key terms and principles in dashboard design
To carry out an idea or question generating exercise the team must have an understanding of the intended users, the raw data, how this data will be used, and general knowledge of the dashboard design domain. To assist this process, we have provided a list of definitions, key terms and principles below that, in part, represent some of the information necessary for all dashboard designers.
Dashboard – A dashboard is a type of user interface specific to an individual, group of people (e.g. employees), institution, department or place of work, which provides graphical representation of raw data in a simple or easy-to-consume form. Information dashboards are used in place of raw data as the viewer can identify patterns, trends or important individual statistics or occurrences without the need for lengthy calculation, analysis or interpretation.
Business Intelligence Dashboard – Business intelligence dashboards are used to show users key performance indicators for their particular business. For example, a business intelligence dashboard for a paper company may show the number of units sold over a specified period of time to help the sales team identify how well they have performed or how business has been impacted by changes in the firm, such as department restructuring or a new advertising campaign.
Consistency and Standards – A key principle in the design of all user interfaces; consistency and standards are necessary as they allow users to apply information and knowledge from one instance to another within the same design. Inconsistent design forces the user to learn new representations and methods of interaction for the same or similar features, which takes more time and burdens them with the responsibility of having to actively remember which action or representation corresponds to each particular instance.
Negative Transfer – Inconsistent design can cause the user to apply inappropriate actions or knowledge from one instance to another. For example, if two bar charts share the same colour palette, but for one the colours are meaningful and in the other they do not represent the same meaning, the user may make inappropriate conclusions when switching between the two. For this reason, consistent design is an essential quality of all dashboard designs and all efforts must be made to prevent negative transfer from one display method to another.
Informavores – When users open their dashboard, they are seeking information relevant to their current aims and objectives; all other information recedes into the background, receiving minimal or no attention at all. This behaviour has seen users termed ‘informavores’; in the same way carnivores seek out meat and ignore fruits, vegetables, grains and all other foodstuffs, informavores search for task relevant information only.
Information Scent – Informavores follow the scent of information. All information carries a ‘scent’ and the stronger the scent the more likely the user will follow. Therefore, it is important that you design dashboards to draw users’ attention to task relevant information and limit the effect of other information on the user when carrying out their activities.
Fragmentation – Fragmentation refers to the act of splitting information into separate sections, pages or positions within the same display. It is recommended that fragmentation is avoided wherever possible, as this slows down the process of comparing and switching between data sets, graphics, display media or any other feature. One of the main reasons for using dashboards as a data visualisation method is the capacity for quick and easy analysis, interpretation and comparison; by breaking the information into independent and separately accessed regions of the display the user cannot see all of the information at once, which limits the viewing experience as a whole and specifically the potential for valuable insights to be achieved.
Scrolling – Arranging information so that the user has to scroll up and down to move between different items, data sets or any other element impacts the viewing experience and potential for insights in the same way as fragmentation. Therefore, when you are dealing with large amounts of information, placing some out of view can sacrifice the usability of the display. In addition, forcing the user to interact with the display moves them out of the mindset of analysing and consuming the data to working out how to find task relevant information. When the user would benefit from having all information on one screen at the same time, try to use your creative nous to fit everything in before resorting to scrolling.
Short-term Memory – Most people are only able to retain between five and nine individual items in short-term memory for an upper duration of nineteen seconds. For this reason, it is important that the dashboard is arranged in a way that relieves the viewer of having to retain information, this means, wherever possible, all information necessary for a specific activity should be placed in close proximity and without the need for interaction (see also: scrolling and fragmentation). For example, if the user wishes to compare the performance of all employees, they should able to do so from one display method and without the need to switch between data sets.
Data-Ink Ratio – A term coined by Edward Tufte, which calls for designers to only include elements that carry meaning in the display. For every piece of ink used, the meaningfulness of the dashboard should increase in turn. The dashboard is a method of communication; by including meaningless graphics, colours, lines or any other design element, the concentration of meaningful data is reduced, and the user has to discern between data-carrying elements and screen filler. Therefore, Tufte recommends designers should use ‘ink’ (or pixels) sparingly; otherwise the clarity and usefulness of the display may suffer.
Exceptions – As dashboards are often used to visualise active data (i.e. the information is constantly changing), it is necessary to save them the time of identifying which elements have changed and draw the user’s attention to these changes. The design elements that highlight changes from one viewing instance to another are referred to as exceptions. Exceptions are an excellent example of helping users follow the scent of information; by highlighting changes, the user can ignore large sections of the dashboard and focus their attention on specific elements.
Aesthetics and Positive Affect – The look of a display can affect the viewing and user experience; if information is arranged in a logical fashion and according to the various tasks the user wishes to carry out the dashboard will more than likely have aesthetic appeal. This is not a measure of how attractive a neutral viewer rates the display, but an indicator of how well the dashboard has been designed according to the specific needs of the intended users. Positive affect refers to the positive feelings we experience when something appeals to us, whether this is something that makes our working lives easier or satisfies us visually, among many other possible examples. Designers should make the display as aesthetically pleasing as possible, without sacrificing meaning or reducing the data-ink ratio.
Schemata – When users approach any product, be it tangible, virtual, or graphical, they bring a series of expectations with them, which are based on their previous experiences. These expectations are referred to as schemata (singular: schema) and they are important as they impact greatly on the way users extract information and interact with the display. For example, if you have included a meaningless, superficial element that resembles tabs in your dashboard, users will expect they can interact with them, which would be incorrect and frustrating for the user. Therefore, it is often important to take the user’s schemata into consideration when designing dashboards to ensure the display looks and behaves as expected.
Gestalt Laws of Perceptual Organisation
Gestalt Psychology is a theory of mind that suggests the brain is holistic, parallel, and analog, with self-organising tendencies, as opposed to feature driven. For example, when we look at someone we do not perceive individual features first; instead we see a complete face. The Gestalt laws affect the dashboard viewing experience as we tend to perceive items as grouped under certain conditions.
Law of Proximity – A Gestalt law, which identifies the perceptual bias to see items in close proximity as grouped. This can affect the perception of data within a dashboard, as the user may erroneously perceive items as grouped if they are unrelated, but placed close to one another. Therefore, it is important to use ‘white space’ to separate items or data sets that are not related.
Law of Similarity – A Gestalt law, which identifies the perceptual bias to see similar items as grouped. For example, if the display contains a number display methods that use the same colour palette, the user would likely perceive them as connected or related in some way.
Law of Common Region – A Gestalt law, which identifies the perceptual bias to see items as grouped when they share a common region. For example, items falling within the same delineated region are likely to be seen as belonging to or forming the same group.
Colour Vision – Colour is one of the pre-attentive attributes. Pre-attentive processing is the unconscious accumulation of information from the environment and as this processing bypasses the conscious mind it is automatic. When we look upon a multi-colour display, we immediately see the individual colours; this automatic processing can be harnessed to convey meaning in an instant. For example, using red, yellow and green to represent three different states, such as bad, moderate and good, can provide users with an immediate understanding of certain key performance indicators. A simple red or green dot next to values in a table allows the viewer to look down the columns or across the rows to establish the state for each value or data set. Therefore, colour is one of the most important considerations when designing dashboards. Within this consideration are a number of other factors that must be accounted for, such as those with colour vision deficits, the nature of normal colour vision, and colour-based schemata, such as the red, yellow, and green example above.
Fovea Centralis – At the centre of the retina (a region at the back of the human eye) is a layer referred to as the fovea, which contains millions of photoreceptor cells. These photoreceptor cells are responsible for colour vision and due to the variation of concentration of these cells across the fovea, human vision is at its best when light falls on the centre of the fovea. We are constantly moving our eyes to ensure our gaze is centred on the most important information in our environment. For this reason, the fovea and the distribution of photoreceptor cells on this light-sensitive region of the eye is an important design consideration. In essence, important information should be given prominence within the display.
Cone cells – These light sensitive cells within the retina give rise to our perception of colour. There are three types of cone cell, distinguished by their different light sensitivities (i.e. low-, medium-, and long-wavelength cones). They each contain pigment, which undergoes chemical changes when light falls on the cells. Different colours cause excitation of certain cone cells and the electrical signals are sent to the brain via the optic nerve. The three cone cell types form three different channels, which act in opposition to create our experience of colour. For this reason, certain colour combinations can be hard or impossible to perceive. For example, red writing on a green background can be hard to read and different intensities of the same colour can be difficult to tell apart.
Display Media – Every designer has a whole array of display media to choose from when converting raw data into a fully realised, graphical representation. For example, bar charts, pie charts, line graphs, and icons.
Comparison – One of the main reasons for using a single screen visualisation of data is the capacity for comparison. By placing all of the information on the one screen the user can quickly shift their attention from one graphic to another, without having to interact with the display. Dashboards must therefore be designed in a way that supports comparative analysis and limits the need for movement through different display methods, sections of the screen or separate pages.
Insight – By placing data on one screen in a way that supports quick consumption and understanding, the user often gains new insights. For example, business intelligence dashboards often enable employees and business owners to judge performance of certain product categories or consumer behaviour. Insight is often gained by comparison, so arranging the dashboard to support meaningful comparisons, whilst preventing or discouraging meaningless comparisons, is essential.
Drilling Down Methods – One of two broad categories of dashboard interaction. There are times when it is not possible to arrange all of the data on one screen; on these occasions the user needs some means of interacting with the display to allow prompt access. Drilling down methods involve interaction to access options, functions, or data deeper within the dashboard on the single screen. For example, the designer may include tooltips, which are accessed by hovering the cursor over an item, to display the values of each bar in a bar chart. Drilling down methods are useful as they help to preserve the clarity of the single screen, yet users are able to access specific information quickly and with the minimum amount of interaction.
Splitting Methods – This involves splitting information into distinct regions of the dashboard, such as separate pages or tabs. Splitting methods should only be employed when the separate sets of information are unrelated or the user would not gain anything from seeing the data on one screen all at once. For example, different departments within the same business might have separate dashboards, which the users can access by selecting the appropriate option from a dropdown menu.
Usability Testing – As is the case with all products, the designers cannot get everything right on the first attempt. In order to identify areas for potential improvement, the design team must establish requirements and determine what they have got right, and what they have got wrong (obviously there are shades of grey in between). The most common way to do this is to test real users or evaluators (i.e. paid professionals, such as usability experts) with a series of paper prototypes or early scaled down versions of the dashboard. Usability testing with dashboards is usually a simple process, as there are fewer factors to take into consideration than other types of user interface, such as websites (where there are typically lots of different interactive elements and pages to accommodate). Improvements are made during an iterative process – with each implementation stage proceeded by a testing phase to check whether the changes have solved the problems identified previously.