2019 in Review: 5 ways Human-Centred Design has improved Tableau at Wellcome
Over the past year, we’ve been using human-centred design (HCD) principles and methods to improve how we use Tableau for internal analytics. Senior Tableau Analyst Natalie Leach and User Researcher Aoife Spengeman explain how.
Separating the problem from the solution.
Natalie: We were always focussing on the solution (“We just need to build better dashboards!”) and therefore we didn’t understand our problem. Design thinking forced us to look more critically at the problem and by doing research we understood it better — it wasn’t that we didn’t have good dashboards, it’s that people didn’t know they existed. This led us to refocus our strategy on a program of events to raise our profile before we moved onto improving the dashboards.
Aoife: The natural tendency is just to want to “FIX IT” with what seems like the most obvious solution. What I love about human-centred design is that it forces you into the uncomfortable zone of spending time understanding before doing anything.
Taking a broader look at the experience of Tableau as a service to our internal users, we ran a series of workshop where we followed IDEO’s HCD process, which helped the team:
- Agree their most important issues.
- Understand the problem based on evidence and come to a consensus.
- Brainstorm a range of solutions in a creative mindset.
- Prioritise and refine solutions, taking action on top ideas.
- Continually test assumptions.
Trusting the process and learning to love finding problems
Natalie: The first time I watched usability testing and witnessed a user fail over and again at using one of our dashboards I was disheartened. But Aoife kept reminding me “This is good. We want to find problems. Then we work out solutions. Then we test again.” She encouraged me to work through that failure and find the opportunity in it to improve. And in the next round of testing, when our solutions fixed the problems, it felt SO GOOD!
Aoife: Usability is a great place to start when you don’t know where to start improving your product. Usability testing is a behavioural observation research method where developers give us the main tasks that they think their dashboard should be able to do. Simply asking users what they want produces a long list of diverse answers that are near impossible to interpret and figure out a solution to. With usability testing you watch what users do rather than what they say. By seeing where users are going wrong and digging into the problem on the spot with them, defining the problem and finding the right solutions is easier.
Realising Tableau is not a perfect product
Natalie: Some of the usability issues we found couldn’t be fixed by redesigning the dashboard. If you have a dashboard with lots of filters at some point a user will apply so many that there is no data left and the dashboard is blank. This freaks them out and they think they’ve broken it. There is an easy way to fix this: Tableau has a revert button which returns the dashboard to its default state. Except no one has heard of a Revert button. Our solution? Make sure the revert button is covered in all our training materials and education programs and remind people over and over — you can’t break a dashboard!
Aoife: User research is not just useful for improving the design of the dashboard, but for informing how you can improve the overall experience of what it is like for someone to use your product. Erika Hall’s book, “Just Enough Research” has been a great resource for thinking about understanding the user experience holistically https://abookapart.com/products/just-enough-research.
Drawing on interaction design principles
Natalie: Some of Tableau’s bells and whistles must be used with caution. Tableau Action filters can be tricky if people are new to the product. If you don’t indicate that a change is possible or provide enough feedback for when a change occurs (e.g. ‘Click on this bar’, ‘You have filtered by X’) then it can be surprising for the user when they accidentally click on a bar and everything changes.
We also found that its best not to pre-filter the data, even if people say they’re only interested in the Top 10. Yes, it might impact performance to surface all the data but not doing so impacts trust. In one dashboard we had a very clever table. It used Top N calculation and Sets and Parameters and it pre-filtered to show just the biggest and best. For a Tableau nerd it was the bee’s knees. For the user it was a disaster: “Stuff is missing. Where is everything else? I don’t understand it and I don’t trust these numbers.” You might come up with a creative solution using advanced Tableau functionality but if it doesn’t fit with people’s mental models with how a dashboard (or a website) works then it’s confusing for them.
Aoife: Interaction design principles and rules have been around for a long-time but we hadn’t adopted them into our Tableau dashboard building. Doing usability research helped us identify the problems and design principles helped solve them. We used Jakob Nielson’s 10 Usability Heuristics for User Interface Design:
- Visibility of system status
- Match between system and the real world
- User control and freedom
- Consistency and standards
- Error prevention
- Recognition rather than recall
- Flexibility and efficiency of use
- Aesthetic and minimalist design
- Help users recognize, diagnose, and recover from errors
- Help and documentation
Shifting our outlook to human-centred design
Natalie: This approach taught us humility. Aoife always reminded each participant “We are testing the product, not you” and that stuck with me: if there is a problem it’s the dashboards fault, not the persons. And ultimately as the designer of the dashboard that’s on me. If the user can’t use it that’s because I haven’t made how to use it clear. This is basic stuff for designers but generally Tableau analysts are analysts first, designers second. We’re prone to making assumptions about user’s data literacy and ability to parse information in ways that seem obvious to us. This shift helped us to create dashboards that people could quickly and easily understand. We went from people finding a dashboard baffling to being able to immediately articulate what it was showing them as soon as they opened it.
Aoife: One of the reasons I am drawn to the approach of HCD is that it makes good decision-making easier. By putting users at the centre, it eliminates the need for egos and lets evidence bring consensus to teams.
The two questions we keep in mind when building a new dashboard are:
- Who are you building this dashboard for?
- What problem are you solving for them?
I find it liberating: You start with being clear about your assumptions then move to being open to being wrong.
Have you applied human-centred design to Tableau development? We would love to hear about your experiences!
To learn more about human-centred design, here are some resources:
- Nielsen Norman Group
- IDEO (they do a great free course in HCD)
- The Double Diamond
Other useful theories, principles and rules to be aware of are:
