10 Questions to Ask for Data Visualization Preparation

After having completed all the ETL and analytical effort, having prepared calculated measures and fields for in-depth insights about the underlying dataset, the challenge becomes how to communicate these insights in your dashboard. The dashboard is the final product that the end user will be reviewing and interacting with, and the way it is laid out plays a fundamental role in their impression of it and the degree it provides a clear narrative that is actionable by presenting recommendations for decision-making. There is no one definite correct way to create a great dashboard, but there are principles that can be followed and questions you can ask for how to make it as good as possible by taking into account a multitude of factors in the situation. In my previous article I focused on 4 principles for making great visualizations, and I will here supplement those insight by presenting 10 questions to consider in the preparation stage of creating a dashboard.

1. Who is the end user of the visualization, and what will it be used for? Are you making a dashboard as a task in your day job for your manager to make actionable decisions from, or participating in an online data analysis and visualization challenge, or maybe just practicing and want to showcase your abilities to your connections and potential employers? The reason you're creating the dashboard is a very important factor to take into account when you decide how to present your findings. If the purpose is to present it to the manager so that she can get a clear understanding of the insights and consequently make better decisions from it, the focus should be on presenting a clear narrative and making it possible for her to interact with the data and quickly understand the main take-aways from the data. However, if it's for an online challenge or just for showing off your abilities, it may pay off to spend some more time on the visual design to leave a better impression. In any case, the importance here is to not forget the purpose of your efforts, as it plays a big role in how you should set up your dashboard.
2. Should the dashboard be dynamic or static? Would it be better to have a dashboard that's flexible and has several filters to interact with, or a static visualization that shows one specific set of insights? Also this depends mostly on the end user. For challenges where the submission is just supposed to be a static visualization, it isn't as necessary to focus a lot on filters or interactivity, and the most important part is the first impression of the dashboard. In those cases whether it's dynamic or static isn't very important, only that the picture being chosen for the submission is good. If it's for a job, however, it can matter a lot, and where it's to be used as basis for decision-making it's best to leave it flexible to be interacted with (while presenting a clear narrative and recommendations), while where it's to be used for visualizations like posters or advertisements, static one-pagers are the best bet.
3. What size should the layout of the dashboard be? Before you start placing your numbers and graphs into a sketch of the dashboard, you should take some time to consider how you want the dashboard to look. The layout is the first decision to be made here, as it is the canvas upon which the dashboard will be painted. Choosing the size of the dashboard matters to make sure that the desired numbers and graphs will fit, while also not leaving too much free space. Which size to choose depends a lot on its purpose, and whether you find it more fitting to present it in landscape or portrait orientation. In most cases, dashboards are made in landscape orientation, and the most common width-to-height ratio is 16:9, which is the HD format and size most monitors follow. For simple dashboards, a 1280x720 size layout works great, but if the task requires high quality (for instance a high-end client or important challenge), the size can be up to 1920x1080 or even as high as 2560×1440 or 3840×2160 (only where extremely high level of detail and quality is required). Usually 1280x720 and 1920x1080 are the best alternatives for the 16:9 format, however, unless higher quality is required. It still depends on the situation, so feel free to try out different formats and sizes if you want to experiment with what seems the best in your case.
4. How many objects should be included in the visualization? Even if you have many interesting graphs and numbers you want to include in the dashboard, you should be careful about not putting in too many. The more items you put in, the more information the end user will have to process, and this can in many cases confuse the user and distract them from the most important insights in the visualization. It is therefore important to stick to the script and focus on the most important insights, and present the objects to illustrate a clear narrative that is easy for the reader to follow and understand. Although the purpose of the visualization plays an important part, it applies in most circumstances that too much complexity and information will throw the reader off, whether it is to be used for decision-making or for visual appeal. It is therefore important to focus on and emphasize the most important parts.
5. Which colors and font should be used in the dashboard? This may seem like small details, but as I mentioned in my previous article, excellence lies in the details. Making conscious decisions of which color scheme and font to use in your visualization can be a very important detail in making your visualization go from being good to great. A clear example of this is Mina Saad's submission to the Maven Unicorn Challenge, where he focused on specific colors and used the custom font Libre Baskerville. Experimenting with different font types, perhaps even exploring ones you've never used before, can be a good touch to the uniqueness of the dashboard making it stand more out. The conscious use of colors also plays a significant role in being consistent and providing a clear narrative in the visualization. In Stephanie French's submission to the Nobel Prize Challenge, she made consistent use of yellow, grey, and black throughout the visualization, and the simplicity in that color scheme makes it easier to focus on the narrative and the highlighted insights. Having a clear custom color scheme focusing on around 3-5 colors that are suitable for the situation makes for a smoother presentation of the data.
6. How should the background be designed? A dashboard can be good even without giving much thought to the background, but if you want to make it excellent and visually outstanding, it might pay off to put some more time into how you want the background to look. You've already decided the size of the layout and which colors and font to use, so now you just need to figure out how you want to apply the colors to the background. Here you also create the placeholders for your numbers and graphs, so make sure you've completed the analysis process and decided what to include in the visualization for this step. Instead of conducting a lot of trial and failure to get the dashboard to look good by moving around on the objects, making a custom designed background with clear placeholders for where the title and different items will be placed can save a lot of time and end up with a lot better result in the end. If you're an expert at visual design and software like Photoshop, you have an advantage at this step, but you don't need to be a design expert to make great backgrounds for your visualizations. Even just using the tools in PowerPoint with all its shapes and icons could be a good framework for customizing your dashboard, and once you've decided the color scheme, font, and how to sort the objects, it's a much easier process to create a good background design.
7. Should any pictures be included in the dashboard? In some cases, pictures may be a good way to illustrate specific insights from the data. As an example, Stephanie presented pictures of the four people who have won the Nobel Prize twice in her visualization, giving a nice touch with visually appealing detail. Another common case includes logos, and usually at least one picture with a logo or similar relevant picture may be included to provide better context for a first impression of what the dashboard is about. Pictures aren't a crucial aspect of a dashboard, but I would advise at least a logo or relevant picture to be included in the dashboard, and that it's a great additional detail to provide relevant pictures of clear insights as long as it doesn't take away focus from the narrative.
8. Are there any icons or other visual tools that can be used to improve the dashboard? Many platforms have various in-built icons that can be included in your dashboard, and this can be a good way to provide extra detail to make it look even better. For example in the Nobel Prize challenge, Stephanie included relevant icons for each of the genres of prizes, and in the Unicorn Company challenge, Mina used icons to illustrate which measures he used beside the key numbers in his visualization. In other words, icons can be great when used right and in moderation, for providing detail that makes the dashboard look more visually appealing given that they are appropriate in the given context. As with pictures, it's important to take note here that it should add to the visual appeal and the narrative, and that it's important not to include them unnecessarily where the application isn't relevant.
9. Does the visualization illicit sufficient visual appeal and compensate for potential complexity? This is related to question 4 in terms of the complexity of the dashboard. An important principle of excellent visualizations is to not make it too complex, as that can make it more messy and unclear for the user the insights that were intended to be communicated. If there are more items and more complex graphs, there is a higher burden to make it more visually attractive to watch and read through it. An example where this is done very well is Nadia Fankhauser's submission to the Maven Unicorn Challenge, where there are a lot of complex items used to visualize the insights, but which still leaves the user impressed with the dashboard and more eager to read more into it. This is not, however, something I would recommend newcomers to the data analysis and visualization field to do, as it takes a lot of professional experience and competence to manage to pull it off, and in this case seemingly strong visual design abilities. I would primarily advise to focus on making simple and good-looking visualizations with a clear narrative, and eventually as you gain more experience, to try more difficult challenges in making more complex presentations if the audience is appropriate for it.
10. How can a clear narrative be communicated to make the insights easy to understand for the end user? By now you have the background in order and populated the dashboard with the visual objects you've chosen to focus on, so now the final touch is to use it to tell a story. Here it's important to be brief and emphasize the most important points, as too much text can make the entire dashboard more unappealing. In most cases it's best to avoid long paragraphs and focus on the most important things. For my own dashboards, I've kept it to just a few short lines per item and included the items as part of the text, but that is just one style, and this is an area where you can really make your visualization your own with your own style and way of doing it. In some cases, short paragraphs that are concise and focus on the insights in the relevant graphs are great for providing extra detail, such as in Stephanie's visualization with 1-3 lines per paragraph. The text you include in your dashboard is your way of communicating more explicitly the narrative you want to present of the underlying insights, so make sure to put some extra time to consider this in order to optimize your presentation.

There may be more questions that can be considered when preparing for and creating an excellent dashboard, but I think these are some of the most important ones to take into account. Let me know if there are any other questions you would like to add, and feel free to contact me to request advice or feedback on your dashboards and how they may be improved.