Building your data visualization creation process
In today’s scenario, creating dashboards and data visualizations has become a fundamental task for analysts and developers. However, without a structured process, it is easy to get lost amidst so many possibilities and tools, resulting in visualizations that do not meet users’ needs or do not communicate information clearly and effectively.
Because of this, in the last few weeks, I have written a few articles about different data visualization creation processes. All of them are extremely interesting and, in addition to teaching us something, also speak to other creation processes. But believe me: there are many other processes for creating charts and dashboards besides the processes explained by Andy Kirk (here), Lisa Charlotte Muth (here), and Ben Fry. In the future, I plan to present other data visualization processes.
On the other hand, as good as the processes presented by the authors are, we know that each company, each department, and each developer has their own ways of working. This means that, often, we study the processes of other authors and, in practice, end up creating our own process based on some variations and/or modifications of the processes we have learned. So, here are some questions for you: what is the data creation process that you use? Don’t you have one? Or do you have one, but want to improve your process? In this text, I intend to help you create one.
Steps for Building a Dashboard
The first thing we should think about is: What steps should a data visualization process take? Throughout my career, both academic and corporate, I have seen countless steps, but we can summarize them all in some way as follows:
1. Prerequisites Gathering and Scope Definition: This step involves gathering all the information that is essential for data analysis and visualization. This includes everything from user needs to data sources. With the prerequisites in hand, the next step is to define the scope of the project. This means determining what will and will not be included in the dashboard, storytelling, and analytical report. A well-defined scope helps to avoid “creepy features” and keeps the project focused.
2. Descriptive Analysis, Understanding the Data, and Gathering Insights: Before you start creating the dashboard, it is essential to understand the data you will be working with. This involves descriptive analysis to identify patterns, trends, and possible outliers, as well as understanding the data and gathering insights regarding possible data analyses.
3. Ideation and Prototyping: In ideation, you begin to sketch out the first ideas for visualizations, given the prerequisites and the defined scope (i.e., the information needs of the users). This can include drafts of graphs, layouts, and even the selection of cores and fonts. Ideation should be a collaborative process, involving initial feedback from users and stakeholders. This ensures that the product will meet the user’s needs. Next comes prototyping — a stage in which preliminary versions of the product are developed to test different approaches and obtain practical feedback before final development.
4. Development: With the prototype validated, the next step is to develop the final dashboard. This involves everything from acquiring real data to building the dataset and visualizations, integrating the data, developing design solutions, and implementing the easiest features.
5. QA (or Validation, Iteration, and Continuous Improvement): After development, the dashboard must have its data validated to ensure that it is coherent, consistent, and credible, and it must also be tested with users to ensure that it meets their needs and works as expected. This step may involve adjustments and refinements. It is also worth remembering that, even after launch, the dashboard improvement process does not end: it is important to continue collecting feedback and reviewing the dashboard to ensure that it remains useful and relevant over time.
Finally, it is worth noting that not all construction processes (much less the ones you develop for yourself) must have all of the above steps or must contain steps only related to those mentioned above. For example: in many companies, the ones who develop the dashboard are the ones who will use it, so there are no prototyping steps; On the other hand, in cases where there is a dedicated team to develop the dashboard, it is not always possible to have contact with the areas that will consume the dashboard — so it is not always possible to come up with ideas for design solutions. In other words, there is no right or wrong. There is only a choice of those steps that make sense in your construction process.
Information to be collected
The first step in creating a dashboard is to gather detailed information needed for the final product to be useful and effective — whether done by you or by a product manager, for example. Below, I list some of the most important pieces of information to gather:
User’s Informational Needs: Understanding what users expect from the visualization is the first step to ensuring that the dashboard will meet their needs. This understanding can come through interaction with users — through interviews or written means — or even through questionnaires, completed forms, or indirectly, through a product manager. In addition, each user has a unique way of interacting with data and performing analyses (what we call the ‘analytics journey’). Understanding this journey — from data collection to decision-making — allows the dashboard to be effective.
Objectives, Purpose, and Users: It is important to clearly define the purpose and the main objective of the dashboard. What is the purpose of the developed tool? This understanding of the purpose helps to clearly define the objectives of the dashboard. This will not only guide the design but will also help define the success of the project. Additionally, identify all direct and indirect stakeholders who will use the dashboard. This helps ensure that the needs of all stakeholders are considered in the design process.
Dimensions, Metrics/Measurements, Indicators, and Goals: Just as important as defining the objective and informational needs is identifying the metrics and dimensions that are most important to users. This includes understanding what the key performance indicators (KPIs) are and how they relate to the analysis dimensions, as well as defining the goals for those KPIs. A good objective should be specific, measurable, achievable, relevant, and time-bound (SMART).
Types of Analysis Needed: Based on the objectives and informational needs of the users, and the metrics and dimensions, it is important to clearly define what types of analysis need to be performed with the dashboard. This will guide the selection of the most appropriate visualizations and the layout of the dashboard. The main types of graphical analysis can be found on the following websites: From-data-to-viz , SQLBI Visual Reference Guide, and the Financial Times Visual Dictionary. In the next article, I will write more about the different types of visual analysis.
Data-Based Actions and Decisions: It is crucial to understand what decisions or actions users need to take based on the information presented. This helps define how the information should be organized and highlighted in the dashboard, as well as helps to develop actionable dashboards — unlike those dashboards that Steve Wexler — Data Revelations calls “So What” because they present data but the user, when looking at the data, does not know what to do.
Supplementary Data: Consider what additional data could enrich the analysis. This can include external data or information that is not yet available but could be collected in the future. It can also include information about how users will access the dashboard — whether on desktop, mobile devices, or other means, this directly influences responsive design and usability — how often data will be updated, where the data comes from, and how the data is processed and purged, among many other things.
Data Visualization Canvas
I won’t explain it in this article, but the process I currently use is an adaptation I created together with Adonis Franco, based on a tool that was introduced to me by João Critis and Anderson Paes. Below, I’ll provide an image of it.
Final thoughts…
A structured dashboard creation process is essential to ensure that data visualizations are effective, useful, and impactful. By building a process that ensures the collection of the information indicated above and steps that, in one way or another, allow you to understand the needs of users, and develop, and validate the data later, you will be better prepared to create dashboards that really make a difference. Now, you tell me: what is your development process like?