What kind of Aussie are you? How we used data and design to shake off political stereotypes
Walkley Magazine editor’s note: Fairfax Media’s Political Personas project began with two academic partners on opposite sides of the world. They surveyed a representative sample of 2600 Australians and then spent months analysing data clusters. The hope was to highlight the new fault lines of Australia’s political landscape. Readers first encountered it with a quiz that felt far from academic: “What kind of Aussie are you?” For Issue 88, Fairfax team members Conal Hanna, Matthew Absolom-Wong, Matt Davidson and Andy Ball talk about how they designed the project.
Getting to readers
Conal Hanna, Head of Digital Channels: How do you reduce three months’ work down to 10 words or less?
Such is the challenge of journalism in the digital age.
Pitched in a battle for attention against an exhausting (but seemingly never exhaustive) list of daily political controversies, social media skirmishes, reality TV recaps and the master of all three, Donald Trump, how do you sufficiently arm your long-term journalistic enterprise to compete?
The Political Persona Project was perhaps the most ambitious piece of journalism I’ve worked on.
But what use was that if we couldn’t get people to read it?
Be it on a homepage or in a social media feed, online journalism lives and dies on its “sell”. In the worst cases, that compromises quality. We wanted quality and quantity.
After Brexit, Trump and our own election, there seemed little doubt the political world was changing. But how to document what effect that was having on Australia?
And then how to make it meaningful to people?
We began with the introduction. We knew we needed to make it distinctive and accessible if it was going to resonate with a wide audience. What better way than to tap into readers’ own self-interest? Don’t just tell me about the new political tribes — show me where I sit among them.
Bringing the seven personas to life from data was mind-bending work — immensely challenging, but rewarding too, like how I imagine a novelist would feel creating a character. The difference was we had data to guide the way. We compiled lists of statements that “cluster one” agreed and disagreed on, then tried to interpret their persona from there. Who is this person? What are they like? What could we call such a group?
Involving multiple people in the newsroom helped a lot in coming up with titles. We were adamant that none of the persona descriptions should be pejorative, which was harder than you might think. Harder still was not falling back on old stereotypes. We constantly had to double check our interpretations against the data, to ensure we hadn’t been distracted by our own preconceived notions. There were many surprises along the way, such as our Anti-Establishment Firebrands — who were just as likely to be wealthy city dwellers as they were battlers from the regions.
Social media was a consideration right from the get-go. We wanted people to share their result — but we had to bring our descriptions to life. That was where Matt Davidson’s amazing illustrations came in. (See “building the interactive”.)
Once the quiz was built, the next task was to plan and coordinate a story rollout across five capital cities and regional publications for both print and digital. We were keen for the interactive to launch at lunchtime — still peak for our online audience — and so timed a soft release for that morning’s papers, which allowed us time to test the tool in a live environment before promoting it heavily later in the day, and then drip feeding stories throughout the next week. We’ve generally found that tools embedded in articles are far easier to promote on an ongoing basis. Placing it full width on an index page looks better, but after a while it becomes hard to promote without appearing stale. By embedding in articles, you can use the content to funnel people back towards it.
Working the angles was paramount. Each story was put through the “hook” test: what is this story when reduced to one line? A story on the unexpectedly high level of support to manufacture more in Australia became, ‘What do Donald Trump, Pauline Hanson, Nick Xenophon and 80 per cent of Australians have in common?’ Reporting on the results of the survey by demographic became, ‘How income, age and education affect your world view’. And so on.
Having “case studies” — real people who identified with the theoretical personas — was crucial, although the approach to these is very different in digital. Rather than a print sidebar, these are seen as a critical part of the main trunk story — even if they don’t lead it, the associated images are crucial to a good homepage or social sell.
It might seem forced, but all of these strategies helped take our tool to as wide an audience as possible, which is what you want when you’re working on any project you believe in. Yes, it’s easy to despair or complain about click bait taking over the world. But if you accept the age of distraction as the new reality, the challenge — and it is challenging, but also rewarding — is for journalists to produce meaningful work that can compete in that environment.
Designing the interactive
Matthew Absalom-Wong, Creative Director: The Personas project employed the same technical platform as last year’s YourVote pre-election calculator, but it required a very different approach. Plotting opinions against discrete political parties was one thing, categorising our broad audience was something else. Too dry and clinical and this could appear a cold, psychological assessment. It had to have broader appeal. We wanted to playfully engage the audience, entice them a little, play on their curiosity and yes, their narcissism too. After all, 40-odd questions was a big ask.
Illustrations seemed a great answer to the brief — playful, engaging, fun and unique. It wasn’t until we put pencil to paper that we realised this was no small feat. Seven personas was a good number, but who’s going to identify with a persona that’s a different gender, age or skin colour to them? Multiplied by three age groups, three skin tones and two genders, and our seven personas became potentially 126 illustrations.
Matt Davidson developed a “modular” illustration process, swapping out hair, skin tones and accessories across the genders and age groups. This gave us the flexibility we needed to produce the full array of faces across all personas efficiently. They became the face of the project — an identifiable feature used across all promotional material, print and online.
Matt Davidson, illustrator: The challenge for my part was to draw people with a bit of character that matched their demographics without falling into stereotype. The problem with that is cartoons inherently deal with stereotypes to communicate. I erred on the side of simplicity — but didn’t want emojis! The simple distinctive faces of Hergé’s drawings (the creator of Tintin) kept springing to mind.
Initially briefed to draw seven tribal clusters (active egalitarians, progressive cosmopolitans, sensible savers, ambitious consumers, moderate conservatives, passionate traditionalists, disillusioned battlers) with 18 illustrations for each tribe. Job soon done, it was now up to our brilliant coders to figure out how to make the thing work. Or so I thought.
After further scrutiny of the data, it seemed the clusters needed some tweaks. Tribe names changed and some traits were reshuffled. I found myself pulling certain drawn features from one tribe and merging them with others to match the newly recognised clusters.
I guess we all started off with some assumptions (maybe even prejudices), but as the data was examined and revisited those assumptions were not always supported. If we failed to follow what the data told us, we’d be in la la land, and we certainly wouldn’t have an engaged audience.
Andy Ball, lead developer: The team had many discussions about the sequence of the questions and decided to break them into groupings and experiment with their order. We added analytics events to the different sets of questions so we could track user progress through the interactive and use the data to make an informed decision about what approach worked best.
At first glance working with a team on the other side of the world could be seen as a hurdle. But this was our second time working with Kieskompas and Luminum Solutions, and our two teams quickly fell into a familiar flow. One side would develop and pass on, then, as the first team slept, the other side would kick in — receive, develop, pass on and respond. The 10-hour time difference afforded a nearly 24-hour production cycle. Slack was the communication method of choice.
In an early prototype of the interactive, the persona was calculated on the Dutch server and the response was sent back to the user. However we thought this added an unnecessary delay of a few seconds for the reader. In the end we decided to do all the calculations inside the user’s web browser (on the client-side). This proved to be a good move as it meant the app was totally self-contained and the persona results were returned instantly.
Fairfax contributors to this piece were Conal Hanna, head of digital channels for the Sydney Morning Herald and the Age; creative director Matthew Absalom-Wong; illustrator Matt Davidson; and interactive developer Andy Ball. See the full project: “What kind of Aussie are you?”