Designing Better Analytics
Product Design Principles for Creating Effective, Efficient and Satisfying Dashboards
With the emergence of data centralisation across many industries, the Analytical Dashboard has become an important layer on top of the data that most businesses have locked away inside their organizations. Technology continues to become ubiquitous, and with that, data capture and structure gives us access to macro trends previously thought out of the realms of possibility let alone reliability but what does this mean for the Designer?
What does this mean for the Designer?
The term ‘Dashboard’ comes directly from the automative component that houses a car’s instruments and dials. This display is fundamentally important to the driver in order to monitor the major functionalities and instruments at a glance. Applying the same logic we can start to build up a hypothesis on the elements of a successful analytics dashboard.
1. Map the jobs to be done
People don’t want a lawnmower, they want to keep the grass short and beautiful.
The very first step in designing successful products is to define the kinds of value that users want to derive from the product. Similar to the lawnmower, user’s don’t just want a dashboard they want to be provided with insightful data from which to derive their decision-making.
At the macro-level, designers should aim to solve for ‘information failure’ by understanding the blind spots that user’s currently endure. By taking those blind spots and grouping them into larger themes, designers are able to formulate an opinion on the types of analytics tools the user really wants.
Even Deeper — Map the questions to be answered
At it’s core, a Dashboard is primarily about finding the answers to a set of questions that the user wants to answer. To truly deliver a valuable product to users, it’s important to discover the outcomes and answers that users expect from visualizations, charts and tables.
The use of a dashboard often maps to specific tactical or strategic tasks (jobs) that users have to perform. Designers should aim to discover some countable set of questions that user’s ask themselves during the completion of the job and build visualizations around those things. Achieving deep understanding of the context of user needs generally sets the course for strong, empathic design.
Scope and Purpose
A great dashboard generally serves a single and specific purpose and allows it’s users to easily understand the major features and patterns of a large dataset. With that said, It’s important to limit the scope and subject of information displayed on any one dashboard. Dashboards with excessive, unscoped data are cumbersome, confusing and intimidating
2. Discoverability is Key
Agnostic to the dashboard’s subject, the purpose of data visualization is to create digestible summaries across large datasets. As a result, Designers should actively seek to build in not just a surface layer contrast but also a contextual and interactive layer of contrast.
Data alone is not enough
A top executive of a major real-estate company once requested that we implement a ‘WTF button’ on his dashboard.
On the surface, these kinds of requests seem humorous but in actual fact expose a huge blind spot in the design of dashboard products.
It’s often the case that Dashboards act purely as a ‘flag’ to alert the user to an outlying datapoint. That said, merely displaying a metric in isolation doesn’t actually solve for the entirety of the user’s needs. Designers should think deeply about the questions users will ask when they see outlying datapoints and ask themselves: ‘What Next?’.
Designers are well placed to ideate on the next layer of interaction, what happens once the user has digested the information? Where does the user go, Who will they speak to?
Don’t make me think
Inevitably, processing power and data capture will only grow more powerful. We’ll have more data to shape and as a result a major job function for designers will be creating reductive, pragmatic and contextual experiences.
3. Let Users Explore
When Users seek to glean some information from a dashboard, they’ll already have a mental model of the outcome they’re expecting.
Depending on the nature of the dashboard product, it’s likely that some users will consume the information in a reactive manner. Caused by either external stimuli or by the dashboard itself, users will attempt to expose a countable set of datapoints upon which they’ll either monitor or escalate.
It’s at this point where the exposure of the atomic, underlying data becomes critical. Having identified an outlying datapoint, the user now wants to drill into this and understand what happened. Some dashboard products will allow users a method of filtering down to a particular datapoint. However, for true intuitiveness, users should be able to inspect and perform action on outlying datapoints in context.
4. Build in Proactivity
As discussed earlier, we know that some of the utility of a dashboard lies in it’s ability to work proactively with users to facilitate insight. One of the most effective ways to achieve proactivity is to develop mechanisms that allow users to ‘subscribe’ to datapoints that evaluate to a certain condition.
For designers, this means building systems that specifically ask the user what data they care about and to what degree. In creating such mechanisms designers must be mindful of simply creating ‘noisey’ subscriptions or a ‘set and forget’ attitude amongst users.
In Summary
Not so long ago, Designers would only be asked to weigh in on the visualization of data, creating glossy and sometimes gratuitous chart views. With the rise of the Product Design discipline, it seems as though design now belongs at the system level and not at the visualization level.
Next Steps
- Map the jobs to be done
- What are the things users want to accomplish with your dashboard?
- What are the types of questions they ask themselves before using your dashboard
- Limit the scope of your dashboard product to avoid overwhelming and cumbersome UI
2. Promote Discoverability
- Use consistency and contrast to expose only the valuable and important information
- Think deeply about the kinds of follow-up questions users will have and design secondary interactions around those questions
3. Be smart about about the use of visual elements
- Use in-built biases to convey meaning
- Display certain content, like labels, proactivity when users need them
- Contextually group your visualizations by subject
4. Let Users Explore
- Give users the means to explore their data at the low level
- Make outlying data-points easy to query and action
5. Build in Proactivity
- Let your users subscribe to the things they care about
- Alert your users proactively and help them to expose links between outliers and the factors that cause them.
