Communicating Effectively Through Visualisations
Thinking points, tips, and tricks to help you get a better understanding of communicating effectively through graphs and charts
Have you ever gotten stuck trying to decide whether your chart should be a histogram or a line chart? What about a stacked column vs. pie? What happens when you need to add another metric? In this post, I aim to unpack the analysis paralysis that comes with choosing
the right visualisation.
Effective Communication
Effective communication can mean anything from checking yourself before you fly into a rage at your partner because they’ve loaded the
dishwasher incorrectly, to adapting your vocabulary whilst chatting with a child, to being aware of the grimace on your colleagues’ faces as you
try and explain to them the intricacies of that one exception we have for that one client who has those special requirements. Effective
communication involves being able to convey a message clearly and accurately, while also being able to listen actively and respond
appropriately. I think we at Slido are pretty good at being conscious of how we communicate, considering our hybrid, async-first & international environment. However, as always, there is more to learn about how we can improve.
In this post, I’m going to consider the intersection of effective communication and one of my favourite topics, data literacy. That intersection is visualisations. Visualisations are often where the worlds of analytics and The Business meet. They are the final frontier before cold hard metrics step into a world of meaning, reason, and context. Their importance shouldn’t be underestimated — they help us find issues and
opportunities, tell stories, and prove (or disprove) theories. For some of us, they are the first thing we see when we login in the morning and the
last thing we see before we close up. They are the metaphorical dogs in our professional existence — trustworthy, reliable, alerting us to
potential threats, and always by our side — a (wo)man s best friend. So it’s important we get them right — but it’s not always easy. Finding the
best option for getting your message across, bearing in mind all the complexities and nuances that might come with it, is tricky business.
Although I certainly am not able to provide you with all the answers to dispel your visualisation woes, I will try and lay out some interesting
points which I hope will help you on your journey to becoming a visualisation master.
“A bar chart will be fine”
I feel confident that the vast majority at Slido feel comfortable reading
the most common types of charts bar charts and histograms, line
graphs, pie charts etc. But does understanding charts mean that we are
communicating the information within them in the best way? For
example, the chart above, from the Georgia Department of Public
Health shows the counties which were most impacted by Covid cases
over time. The first thing we notice from the chart is that the general
trend over the period is downward. However, if you check the
horizontal axis, you’ll see that the dates aren’t chronological and that
the data is sorted by bar groups, highest to lowest. This is not wrong
per se; it’s possible to think of situations when it’s useful to arrange
your data like this. Yet in this instance, our focus is on the time element
so it would make sense to look at the dates chronologically.
Now let’s consider the same graph, but with the data grouped by
counties and of course, ordered in time. In this version, our instinct is to
first consider the counties separately, to understand how Covid cases
changed over time, and then to compare the trends with other
counties. The grouping of colours also removes some strain on the
eyes, and the label names being nested inside the graph area makes it
very easy to see which colour represents which area. This is just one
example of how shifting the visual focus in a chart can change the way
you instinctively think about the data within it.
Leveraging instinct in visualisations
There are many ways to leverage instinct (our subconscious
understanding of visual cues) in data visualisation. For example,
diverging colours like the graph on the right, give the indication of
separate or distinct categories. Whereas with the shades of blue on the
left we implicitly understand that there is some commonality between
the categories.
If we see an image like the bubble chart below then we understand that the size of the bubble relates to the size of the metric. US deaths from heart disease are larger than from pulmonary disease and so on and so on.
No one has taught us these things explicitly, but they are things we understand and that are important to consider when you’re choosing a visualisation. Here are some other examples of our implicit (or sometimes explicit) understanding:
- Parts-to-whole: a shape split into segments represents 100% (e.g. pie chart) or a complete range of data
- Order: time is ordered chronologically and values sequentially
- Grouping: objects which are grouped together have some sort of relationship
- Polarity & zero values: the intersection of the axes is where zero values belong, and negative values occupy specific places relative to the axes
- Directionality: an upward movement along both axes is positive
- Time: timelines are linear and forward moving
- Colour ranges: colour scales from blue or green to red imply low-to-high, or cool-to-hot (also passive-to-active)
- Colour gradients: colour intensity communicates quantities
When we bear these things in mind it’s easier to become more creative and expressive with our visualisations.
So how can we help ourselves be better communicators? Here are a few very important questions (not an exclusive list!) to ask yourself when
creating a visualisation
- What is the question I’m asking? How will this visualisation help me answer that question? What message am I hoping to convey?
- Who is my audience? What do they already know, if anything, about the data I’m presenting? Do they already have context, or would it be
helpful to add context e.g. that the huge drop in metric x is due to downtime? Should I indicate the downtime on the graph to add
context? - Would it be helpful for me to add descriptions to the visualisation? What title should I choose? And what about the legend, is it clear
enough for the reader? - How do the data points in my visualisation relate to one another? Am I showing parts of a whole, related categories, or are they totally
unrelated? - Is my visualisation easy to read? Are there things overlapping, are there too many colours or are they too garish, is the font big enough, will
people know what the axes mean?
Updating our mental models
To wrap this up, if we are able to spin all of these tips into the perfect visualisation, what benefit would that even have? Apart from the escape
from bar chart fatigue, we would speed up the time in which it takes to get insights. We’d potentially widen and deepen our understanding of
the underlying data, building the same complexity into fewer visualisations. Most interestingly though, in my opinion, would be the potential to reshape our mental models about the business we work in. Images, in our case visualisations, play an important role in reinforcing mental representations of concepts, ideas, and relationships. Images enhance our mental models by providing visual cues that help to organise and structure information in a meaningful way. By rethinking the way in which we design our visualisations, we can potentially unlock previously
unfound insights and nuances in our product and users.
