Designing data-driven experiences
8 tips to empower people through data visualisations
Data is a great source of knowledge that people can use, for instance, to be alerted if a metric goes down, to evaluate their own behaviour, or to argue for a scenario of their choice. Unlike machines, people can perceive the same dataset as either alerting, motivating, convincing or boring — depending on its visualisation and the triggers we design. That’s why at MOBGEN | Accenture Interactive, we help our clients to empower their users by delivering products that visualise data. As a contributor to several project teams that have involved data as driving force of user experiences, I would like to share my learnings on how to have the greatest impact when doing so.
1. Make your data actionable
While computers are good at collecting immense quantities of data to classify and spot interesting patterns, humans are undoubtedly better equipped for selecting what matters to them and making good decisions. Data visualisation is an enabler for people to comprehend the analytics of a context and act upon the gained awareness. So, when designing a data interface, it is crucial for design teams to study user needs and motivations because people are bound to use the insights from data to pursue their goals. Therefore, all data presented through a product should be relevant to these goals.
In the example case of a banking application, it might not be enough to visualise a user’s financial behaviour if he can not make sense of it, or does not know how to change his spending habits in order to reach a saving goal.
Tip #1: In our user research process at MOBGEN, we use personas as a method to outline characteristics of end users, such as goals, barriers and decision criteria. When designing with data, we enrich them with aspects like, ‘which data is our user searching for’, filtering or exchanging with others. We repeatedly ask WHY a user would need something, as this can help to reveal hidden motivations. Following this, we design a product that can support the user in reaching their goal through structure and action triggers.
2. Guide users towards good decisions
Truly powerful dashboards not only provide a status or timeline overview, but they also take a step towards user goals by interpreting data results and triggering them to act. This can be achieved, for instance, through highlighted elements, notifications or colour coding. Besides, this can also add value if the dashboard can provide a human-way of drawing a conclusion from the data.
Smart home, as well as fitness dashboards commonly provide people with a dataset of their past progress. What if they could spot patterns early on and use this information to actually start an intervention? Users could thereby be challenged to take better decisions and improve their consumption habits.
Tip #2: Triggers for action can be designed in many ways, such as challenges, rewards, educative tips, comparisons or error signs. Find a consistent way to communicate triggers to users, for instance, with an avatar who provides personal guidance.
3. Contextualise your data in- and output
Humans are likely to perceive and relate data to their personal situation and environment. For many products, data in- and output should cause minimal user effort and should be aware of the context. Text or voice, tabs or filters, sliders or value fields should all be defined in a logical use flow and positioned on a meaningful scale. Next to this, timing is important. The cycle of use and notifications can establish the difference between an application giving relevant insights or annoying details that feel repetitive. Useful cycles can differ between annual reviews, weekly checkpoints, daily or even real time updates. Furthermore, dashboards can switch between context specific modes, e.g. to only display highly urgent information.
Technicians, extracting data from an application while executing work, will require an output especially designed for their situation. In another case, a business dashboard could give the user a quick glance at the data on-the-go from a mobile device. However, the same user might also want to present the data visualisation in a meeting.
Tip #3.1: With the method of analysing a day in the life of the end user, moments of context transitions and their usability requirements can be mapped out. This journey draws a curve of emotion, as well as an information curve, that explores which data is needed in each moment.
Tip #3.2: The Reframing Method is a great way to deconstruct the context of a product or service and afterwards envision which factors will drive this context to change in the future. Example factors are social/cultural trends, economics, demographics, politics, ecological changes and new technologies. By following this process, designers can ensure that products stay relevant and meaningful for people in our constantly changing world.
4. Let users drill down into data hierarchy
A simplistic data interface allows users to understand its key message at first glance, whereas details are only revealed when needed. Designers therefore must understand what the most valuable piece of information that they want to convey is, and which level of detail is relevant. This hierarchy will shape the information architecture and the data groups and connections will build up the drill down structure. According to the DIKW pyramid, only when raw data is logically structured into information and then synthesised into knowledge, can users eventually gain wisdom. Once you understand how the data is linked, you are able to find the right type of chart that communicates it. For instance by representing either relationships, comparisons, compositions or distributions of data. Further, consistent colour coding helps the user easily recognise positive/negative values, different categories or buttons.
To allow a user to understand a key message, data has to be presented in the right format. A timeline, for example, can indicate a trend on a defined scale, meaning that the user will immediately see a tendency and can decide whether to react. In other cases, the data is best visualised on a map, relating items to locations. You can explore different visualisation types here.
Tip #4: Simplify your data by reducing it to the very essence and organising it in appropriate groups. Check whether a visualisation conveys the key takeaway within 3 seconds, the explanation of the major elements in 30 seconds and reveal optional details in 300 seconds.
Designing interfaces for data is not only about numbers, it’s visualising information in a way that people can make sense of its relations and meaning at a glance.
5. Tell your story through data
When trying to reduce complexity for users, a narrative approach can be helpful to communicate a dataset step-by-step. Thereby, curiosity is generated and the user’s attention is guided through the interface in order to tell a memorable story. Data storytelling can also be a good way to initially explore which message structure people want to understand from the interface. While narrating, it can be very insightful to map out the data model of the back- and frontend of the service. In this activity, data scientists can complement the team by thoroughly analysing a data model. Thereby, information gaps in the story that will need to be closed in future, can be spotted.
Especially when data is managed over long timeframes and many technical stakeholders are involved, a red thread throughout the story of the data can help to align communication in one place, highlight data relations and track key points of interest.
Tip #5: The design team can use a narrative approach to act-out how the data story would be explained, for example to a grandma, child or colleague. Another way is to act as if the data interface would be human, and find out what questions user would ask it. The data dashboard will have to contain answers to these questions. This approach does not have to, but can lead to innovative ideas, like conversational chatbot interfaces.
6. Facilitate a living data canvas
Dashboards not only present data, but are designed for people to interact with data. A living data canvas is an open grid with many points of interaction. Next to this, data animations and transitions are very powerful tools that can bring visualisations to life. Animations illustrate how data dynamicaly adds up over time, gets distributed among categories or compared with each other. To make the most out of mobile dashboards, the use of gestures should be explored. Delightful feedback of touch micro-interactions give mobile applications a human feel.
Dashboards, which let users manipulate the view on data, will be perceived as more immersive due to direct feedback. For example, through interactive filters or sliders that move the calculation. Results of a certain action are clearer to understand, while animated system processes keep the user informed about what’s going on.
Tip #6: In case a dataset needs to be communicated to multiple stakeholders, it can be beneficial to let users choose custom settings at an entry point, so they are presented with only those data widgets that matter to them most.
7. Let users manage their own data
While sensors enter our homes and applications track our health-, financial- and social behaviour, designers need to shape the way we experience our own datasets. True data ownership means that users are able to partially collect data, are in control of what data is recorded and can grant access to their data to services or institutions. Enabling people to make use of their data can give them a great sense of self identity and management.
For example, wallet apps make personal files easy to access. Now, imagine a patient being able to measure data at home, he might either find out what he wants to know through an application, be helped by a remote doctor or be prepared for a consultation. Over his lifetime, he could track his medical history and move around the world while always being able to let any pharmacy access his medication or vaccination list.
Tip #7: In a service blueprint all touchpoints with data can be mapped out to define a user’s experience. Challenging moments can be focused on and the design can try to drive a more effortless, confident or efficient experience.
8. Prototype with data
After gaining an understanding of the data structure to present and the ideal experience to create for users, move your ideas forward by prototyping them. Low-fidelity prototypes work well in initial co-creation sessions with clients or real users involved. This is because all participants are given the tools they need to quickly express their thoughts or critique and the design structure is deliberately left open to profound changes. In this creative process, the team learns early on whether an idea will successfully address the user needs.
In contrast, ill-defined dashboard content can be irritating during user tests. All data in the prototype should be correctly labelled, with sensible examples and units defined and the copy including well-written text in a suitable tone of voice, rather than placeholder text. With high-fidelity prototypes, the navigation through data relationships can be validated. Besides, through testing multiple versions, the most easy to understand visualisation type can be chosen. Also the affordability can be tested, when evaluating whether interactive elements have sufficient use cues.
Finally, prototyping can give valuable insights for implementation. When experiencing a realistic prototype early-on, developers can indicate how difficult or time-intensive building the data visualisation will be. Thereby, the team can iterate efficiently and back-up plans can be made.
Tip #8: Think out of the box when prototyping for data experiences. Start drawing data on transparent foil as overlay for images, to explore how information can be visualised for virtual reality experiences.
Summary
Start by simplifying data dashboards to the essential information that users can understand at a glance and use to act towards their goal. Find human ways to help people interpret data and make good decisions. Think about the context of use and which emotions you want your users to experience. Data is not only about numbers, so choose a meaningful visualisation to convey data in a story and let each individual drill down into details, that matter to them. Interactive or animated data can foster understanding and delight. Always co-create with clients, real users or data science experts, discuss feasibility with developers and test your designs through prototypes.
Conclusions
The key to powerful experiences with data is meaning. Only if people can relate the information to their lives, the dashboard can be successful on the market. There isn’t one right method of designing meaningful data interfaces, but I hope this post inspired you to explore some of them, to figure out what works best for your team.
I’m a Product & Service Designer focussed on user research and prototyping of meaningful experiences. Currently based in Amsterdam, part of the team at MOBGEN | Accenture Interactive.
Hopefully you found this post interesting! Please feel free to share your feedback with me, or say Hello on LinkedIn.
