How can you become a good SAP Analytics Cloud dashboard designer? This is a difficult question because there is no ultimate way to achieve this goal.
Similarly, you could ask: “How do I become a good football player?” The simple answer is learning by doing. Of course, there are a few fundamental rules that you should keep in mind before you start your journey to becoming a dashboard design hero.
Most importantly, you should increase your knowledge about the product itself. Before you try to take your first steps and aim to build a perfect version of your dashboard, get familiar with at least a few basic functionalities that SAP Analytics Cloud provides. Most notably, you should learn how to create different types of charts (e.g., bar charts, variance charts, or just single KPI charts to display numbers) or how to setup a table (e.g., drill-down possibilities, hierarchies, and measures). You don’t need much at the beginning to display basic data information in the application, and when you’ve become an expert in the charting and table area, you’ve already come quite far.
Beyond these functional aspects, you need to familiarize yourself with all the different formatting options to lay out your dashboard. SAP Analytics Cloud offers tons of formatting possibilities. Having a solid overview of them helps you to ultimately fine-tune the dashboard and adjust it according to corporate branding. Of course, from a functional and design point of view, there are almost no boundaries. After you became familiar with the basic functionalities, you might go one step further and use more advanced features to optimize your dashboard, for example, by offering possibilities to the user to let him play around with the data. This can be achieved, for example, by introducing geo maps, waterfall charts, story filters, or hierarchies; everything that can give users the best opportunities to analyze the data.
If you want to learn more about how to best use SAP Analytics Cloud for your business, you should visit www.sapanalytics.cloud/learning/ to find plenty of online courses, videos, tutorials, and webinars. We especially recommend searching though the webinar section of this website because it provides a great overview of upcoming courses where you can also get in contact with a few selected product experts from SAP.
After you’ve familiarized yourself with the functional parts of SAP Analytics Cloud, we recommend taking another sidestep before starting with the design of your first dashboard. Please keep in mind that the design itself usually isn’t the most critical part of any dashboard exercise. Before you even think about what a dashboard should look like, you should first consider the data and the story that you want to tell. If a dashboard doesn’t tell any story to your users, it isn’t worth the time, so you should think about the message you want to send.
Who are the users that you’re engaging with (personas)? What do they need, and what is their task? What do they expect, and which information do they want to get out of the dashboard? When you can answer all these questions, you’ll be ready to build a dashboard prototype based on the information you gathered. Creating a user journey map can help identify concrete use cases. We’ve seen many dashboards that were created before the dashboard builder considered both the use case and personas as well as the user journey. What usually happens is that most of these dashboards are a waste and need to be rebuilt, which takes time and costs money. That is why careful preparation is indispensable.
Customers often ask the following: Do we need designers to create great dashboards? If you aren’t a designer, it might be difficult at the beginning to create nice dashboards for users. However, the answer to the customers’ question is clearly “no.” In past years, many of our internal dashboards were created by people without a design background or without having studied in a design-related field. It doesn’t matter if you’re a data scientist, business analyst, finance manager, or something else. That’s the beauty of dashboard design: everybody can learn it, independent of their profession.
However, that doesn’t mean that a dashboard could not be improved by the help of a designer. After you’ve created a dashboard, it can make sense to get some help from an experienced design colleague. This has two advantages. First, you get the chance to improve your dashboard with the professional knowledge of a colleague who has a design mindset. Second, you get the chance to learn more about the design skills needed and how to best leverage those to improve your dashboard. Thus, including a few review sessions with people from the field will help you again master your targets.
When it comes to dashboard design, there are ten golden rules, set up in two parts, according to the design principle of “form follows function.”
The first five rules ensure that appropriate visualization types are chosen. The next five rules ensure that the form of the dashboard is optimized to give confidence for decision-making. Below lays out the complete list of the rules.
Let’s now jump into the rules:
The simplest form to store and read data is a table, a container storing the data of two or more related data sets in columns and rows. The information is encoded as text, including numbers and words.
Data sets can be dimensions or measures. Usually, tables contain measures, but sometimes they show dimensions only, such as the type of service level agreement for a list of customers. With measures, tables usually display detail values, but also summary values, such as a total. Tables make it easy to look up and read individual values precisely. Numeric points can be seen as the smallest possible tables, containing just one cell. A heatmap highlights the biggest values and makes it easier to recognize patterns in large tables.
Some examples of where this applies are incidents per age group over time, and a list of customers to be called most urgently.
Comparing measures of magnitude is the most common purpose for dashboard consumption. While tables only support the comparison of two measure values, bar charts are made for comparing multiple values at a glance, especially when ordered by rank, vertically arranged to show changes over time and horizontally arranged for other categories. Watch out for a truthful scale to prevent Tufte’s lie factor. A variance chart on top allows you to display yet another comparison, such as last year or plan data (see also IBCS; https://www.ibcs.com). A line or area chart is often used to compare values over time.
Some examples of where this applies are sales per region or stock price over time.
A number only has meaning in a given context. A tree map presents the number in relation to other values as part of a whole. A stacked bar chart works like a mini tree map. Pie charts are difficult to compare and eat up lots of real estate on your screen. Waterfall charts reveal increases or decreases between periods or categories. Gantt charts show the distribution of work per period.
An example of where this applies is a budget breakdown.
Data is often bound to a certain place, such as the five continents on earth. A geographical bubble chart highlights the aggregated summary of values related to a certain area in proportion to other areas on the map. Watch out for misleading screen space real estate (lie factor), for example, when showing election results from huge areas with little population in a geo choropleth map.
Some examples of where this applies are population density and cycling injuries in traffic.
Other often-needed chart types can be summarized under correlation and distribution. Correlation is the mutual relationship between two or more measures. The scatter plot serves to display this relationship, supporting the analysis of values from two measures. Bubble charts allow you to visualize correlations between three measures.
Charts that are well suited for analyzing the distribution of a set of measures are the box plot and the histogram. The box plot shows how data in a sample distributes around the median, using indicators such as lower quartile and upper quartile that form the box, the highest and lowest value that form the whiskers, and sometimes also outliers. The histogram is a bar chart that clusters numerical values into bins and displays their frequency.
Some examples of where this applies are outside temperature and ice cream consumption, analysis of the distribution of usage tracking data, or number of plane arrivals per hour of the day.
A design grid is a series of intersecting vertical and horizontal lines to build a framework to align graphic elements on a page in an easy-to-grasp manner. Newspapers are composed of such a grid.
An example of where this applies are small multiples and a series of charts using the same scale and axis arranged in a grid to support gestalt laws such as continuation, similarity and proximity.
The visible light spectrum for the human eye spans from ~700 nm (red) to ~400 nm (violet). Adding up all primary light colors of red, green, and blue leads to white; the absence of light is black. Similarly, complementary colors (red-cyan, green-magenta, blue-yellow) are most contrasting and cancel each other out.
Some examples of where this applies are shades (mixing a color with black) or tints (mixing a color with white), which can be used to define a palette of distinguishable tones of the same color when working with heatmaps or choropleth maps.
Keep the design simple and neutral. When applying the preceding rules, your dashboard already inherits differentiating readability features such as small and big bars or dark and light color tones.
Because every dashboard is made for decision support, it should be designed as simple and neutral as possible. You may achieve this by omitting unnecessary elements and redundancy to optimize Tufte’s data-ink ratio and picking an unobtrusive color that is either grayscale or in-brand for main graphic elements such as type and charts. Additionally, alerts or highlighted comments should be visible at first glance.
Some examples of where this applies are deemphasized user interface (UI) controls. Expose negative/positive variances in red/green and highlight key messages via comments in a contrasting color with no semantic association.
A uniform style enhances the readability and reduces the time to understand and act on the messages and patterns detected. Fonts, colors, and imagery should adhere to a corporate style guide. Check also for integrity: Do your charts have truthful titles, legends, scales, and unit of measures? Did you watch out for logical flaws? Apply the Minto MECE Principle to separate mutually exclusive and collectively exhaustive (MECE) information.
Some examples of where this applies are when showing sales data for France, Denmark, and Germany (40, 20, 30), adding up to a total of 100, but forgetting about the “others” category (10 accumulated in smaller countries).
Think of the experience you foresee for the dashboard users, such as giving a talk aiming to convince the audience, participating in a vivid debate aiming to make the best decision possible, or exploring an unknown data set to reveal a story. Ask yourself why they care about the numbers and what led to the peaks and insights buried within. A dashboard can be a communication assistant, common knowledge board to make decisions on, or an information management system fostering curiosity and creativity for people working with data.
Some examples of where this applies are consistent and easy-to-operate action placement for input controls such as filters. Use the peak-end rule: design first and foremost for the most exciting and final moments of your user’s journey.
The figure below summarizes all ten of the golden rules.
If you want to stick to a common look and feel for future dashboards, it helps to define your own design standards and guidelines. Those standards and guidelines can help you do the following:
Thus, our recommended approach is as follows: First, create the dashboard based on your user journey maps and personas. Second, provide a set of rules (standards) that other designers can use for their own work. These rules can incorporate any UI-related information, such as the color and size of the fonts and headlines, spaces between different widgets, types of charts, or the number of KPIs shown on one page.
If you’ve familiarized yourself with the product, invested time to understand the use cases and personas, created a user journey map, and learned from your design colleagues about some design best practices, nothing can stop you from building your first dashboard. Now, you have all the ingredients together and can make use of your skills.
Remember that the best way of becoming a dashboard designer is learning by doing. Nothing is better than experience in the field. Don’t get disappointed when your first steps take time and your first dashboards don’t seem to look as professional as you would hope. That’s natural and part of the game. As in every discipline, don’t give up, but continuously learn from others on how to improve. You’ll see that your second dashboard is much better than your first one, and the third one is much better than the second one.
Good dashboard designers aren’t born the way they are, but instead they invested a lot of time to familiarize themselves with the subject. In the end, becoming a better dashboard designer is only possible if you incorporate the feedback from others and learn from it.
Learn how to build an SAP Analytics Cloud dashboard with SAP S/4HANA Cloud data here.
Editor’s note: This post has been adapted from a section of the book Designing Dashboards with SAP Analytics Cloud by Erik Bertram, James Charlton, Nina Hollender, Melanie Holzapfel, Nico Licht, and Carmen Paduraru.