What does ‘design for your audience’ mean?
As has been explained, the focus of our research study is towards understanding the visualisation literacy of ‘everyday’ people and their inherent capability to read, understand and get the most out of data visualisations: What difficulties do they encounter? What type of visualisation do they find easier to work with? How intuitive do they find different techniques?
One of the oft-repeated mainstays of data visualisation advice is to ‘design for your audience’. As a key component of preparatory thinking, you see it mentioned in any worthwhile presentation of good practice.
At the end of this study we hope to further equip the design community with a greater appreciation of some of the key issues about this relationship between a visualisation and its audience.
But what does ‘design for your audience’ actually mean? I have often wondered if this rather tweet-sized soundbite of wisdom is entirely understood by practitioners looking to fine-tune their visualisation design skills.
As we’re trying to help reduce the barriers that might exist between a designer, a visualisation and its eventual audience, I thought I’d briefly offer a few suggestions of the key factors that I try to take into consideration when working on a visualisation.
Note: See the bottom of this article for comment about the term ‘audience’.
When we talk about ‘designing for your audience’ we are using the language of ‘usability design’ or, more specifically, the sensibility of ‘user-centred design’. There are many definitions flying about out there but let’s consult the gospel, Wikipedia, for one suggestion:
“User-centered design (UCD) is a process (not restricted to interfaces or technologies) in which the needs, wants, and limitations of end users of a product, service or process are given extensive attention at each stage of the design process.”
In a nutshell, this is about having empathy towards our intended audience. A data visualisation should not be designed for you but for others. You are not the best representation of your intended eventual audience. You might be a reasonably close proxy, at best, but it is too lazy to just rely on your own instincts and be driven by your own tastes.
Who is your audience?
What might they know and not know about your given subject? What is their likely level of interest/engagement: how captivated or indifferent to this subject are they likely to be? Will we need to ‘seduce’ them in some way or are they already ‘onboard’?
If the audience is large, varied and indefinable, what is the best attempt you can make to form a typical persona resembling the likely characteristics of somebody who might consume this work? Try to then make your design decisions with these personas in mind throughout.
Familiarity with representation methods
The familiarity and confidence (or otherwise) with the possible representation methods (read ‘chart types’) you might employ will be a critical consideration. Whilst the majority of one’s audience may be adept at working with the Three Amigos (the bar, line and pie chart), there may be other chart types you need to use for the given data and portrayal you are intending.
Context is always everything in visualisation. Whilst we should take into account the possible lack of familiarity with different chart types we should not be impeded by it. One of my favourite recent works by Bloomberg, a connected scatterplot, was a very different chart type, I’m sure, to many within the eventual audience. However, it was an entirely appropriate representation method. Rather than decide to avoid portraying the data in this way they stuck to their guns but did their utmost to provide sufficient assistance in the form of a pop-up ‘How to Read’ guide.
What is the nature of the audience setting?
Will this work be consumed in a situation where there is limited attention-span, time pressures and a shortage of patience? Maybe there will be no scope for the audience having to ‘work things out’ or spend minutes interrogating different parameters to get to an answer. They might be expecting big neon-lit headlines helping them take away the key insights within 5 seconds. This is probably not a time to risk trying out ‘new things’.
Conversely, is there scope for a prolonged exploration? Perhaps the setting is not as high-pressured, perhaps there exists a desire to dive into every detail, nuance and dimension of the subject in depth. Perhaps we can try different representations, perhaps we can provide interactive options to offer many permutations and interrogations.
How will it be consumed?
If you are there to present the work in person, you can provide an explanation of the subject, what treatment has been applied to the data, what the colour keys relate to, what is a good shape and a bad shape, what are the limits to the interpretations etc.
If you are not physically there in person, all that explanation, coaching and audience assistance must be included within the work itself. Being remote from our eventual audience sharpens the need to ensure everything is as understandable as possible for it to survive in the wild away from our supervision.
Are you producing for print or for web? Will it be a large poster or a small graphic in a journal? Will it be for desktop, tablet, smart phone or all the above? Depending on the intended output format you need to carefully think through the opportunities and limitations each will offer. Will the screen of a smartphone be sufficient to include all the detail? Will somebody be able to read the tiny font size of your labelling if it only given a very small part of a printed report?
What do you want people to come away with? Is it about facilitating an exploratory discovery process or putting on a plate the key findings you have unearthed from your subject analysis?
What range of data questions will they want answers to? Will they want to see how something’s changed over time or how it looks spatially? For judgment contexts will they need comparisons against last year or the year-to-date average? We don’t always just have to second guess the needs of the audience for this, our own domain knowledge and our own analytical discoveries should also shape the agenda. But we should give due consideration to what interrogations of the data we think the audience will find useful.
Pre- and post- launch evaluation is something we should do, testing out the proposed and eventual design solutions with representatives of our audience and responding to/learning from the feedback.
This is an ideal. The reality is somewhat different. Testing is an area I would imagine most of the folks in the visualisation field would accept we could do more of and be better at: My instincts and anecdotal evidence suggests it is undertaken less robustly and less religiously in comparison with other branches of design.
A final comment about ‘Audience’
Audience? User? Reader? Consumer? Recipient? There are limitations with all these terms but a discussion about the merits of the various nomenclature is for another day and another post. For this article I’ve stuck with audience but I’m sure other terms would be more representative and accurate. Also, for the purpose of clarity, the focus of this article is about data visualisations as a communication device for consumption by others, not data visualisation as a technique for conducting analysis for oneself (ie. when there is no other audience besides yourself).