Data Visualisation & Storytelling - 5 tips!
Updated: Oct 16
So you have collected your data, analysed it and you think you have a story so now what?
You need to convey your story to stakeholders so they can be informed and if necessary make data driven decisions. Here are my tips with inspiration from Cole Knaflic's 'storytelling with data'.
Tip #1 - Understand the context
When reporting we often show data which is explanatory and not exploratory as this informs the reader and not overwhelms the, You have to consider the following:
Your audience: Are they board members pressed for time, is it a marketing team that needs to understand A/B email testing results or is it a facilities team at a football club needing to decide what to change for their fans for the upcoming season? Whatever the audience, language, tone and visuals need to be considered accordingly. Have more than 1 audience, consider 2 different communications!
You: Are you communicating to a new audience or a familiar one. Are you an expert or will you need to establish credibility? This may impact the flow of your story.
What: What do you need your audience to know or to do? Remember you are the subject matter expert as you have uncovered new insights, data or analysis. If you can't recommend an action, recommend a discussion! Put the question to your audience as reporting/presentations in my view are 2-way and not one-way. If you're not generating interest or excitement with your audience then the message isn't hitting home. Use 'Action' words like ' persuade' and 'promote' and 'respond' to encourage actionable insights!
Mechanism: Live presentation? Great not everything will need to be on the slide as you can convey extra points and answer questions as you go along. Written document? The devil is in the detail and any charts included have to be clear. PowerPoint Report? Visualisations are key but more supporting text may be needed to fully explain the key message. Refining the chart makes conveying the story easier.
Tip #2 - Choose an effective visual!
Simplifying complex data sets through visualisation allows viewers to grasp the main findings quickly and effortlessly. Here are my thoughts with some inspiration from Cole Knaflic's 'storytelling with data'.
Bar charts are simple so use them! It is easy to compare the end points of bars and to quickly understand which categories are biggest (although I like to improve this further by ordering them from largest to smallest from left to right unless the message dictates otherwise!). Make sure that your axis, whatever the measurement, starts at 0 as otherwise this can mislead your audience
No 3D charts - this might be fun to select in a PowerPoint slide but they do not assist you in telling a story about the data. They're hard to read, so avoid!
A lot of data users will also avoid pie charts/doughnut charts (otherwise known as 'dessert visuals'). However, I like to use them but only where there are 2 data points and no more than that otherwise you won't be able to see clearly the proportions. You can alternatively not use a chart at all and use simple text: "20% are struggling with the cost of living." Text is in itself a visualisation!
Line graphs are great for continuous data (data over time).
Scatterplots can be useful for showing the relationship between two things.
Slopegraphs can be useful when you have two time periods or points of comparison and want to quickly show relative increases or decreases between various categories.
Stacked horizontal bar charts are great for Likert/agreement scale based questions. You get a consistent baseline on both the far left and the far right allowing for easier comparisons.
PowerPoint over R, Python and even Tableau. I use PowerPoint for data visualisation because a) charts are easier to manipulate and b) better design quality. I use R and Python for charts but only when I am trying to understand statistics, clustering or test results. And although chart colours, fonts and design can be manipulated, I don't believe they translate as well or as cleanly as they do in PowerPoint. Now Tableau is great for data exploration and data dashboards. But some of the charts you can use on there are quite difficult to comprehend and not appropriate for market research reporting (I am exploring Displayr at the moment which could be the solution here).
Tip #3 - Eliminate clutter
Simplicity and clarity are key when designing data visualizations. Visual elements that take up space but not understanding need to go! Remove lines behind the chart and borders for example. Shorten labels from 'January' to 'Jan' Also get comfortable with white space!
Use clean and concise labels, titles, and legends. Always include your keys on your charts (upper top left) as that where your audience will go to first. Then include the base size and the question put to respondents/participants for context - this will reassure your audience that the data is robust and the context the data sits within ( I usually place those at the bottom of a slide). You don't have to label every data point though, only ones that you want to draw attention to. Ensure that the key message or insights are easily discernible at a glance. Don't rush for that ever so complicated 'bubble' chart!
Your audience is key when it comes to visualisation. You just have to refer to the slides delivered to the UK population on Coronavirus trends. They were messy, clunky and it was unclear what the main takeouts were: “This is a complicated slide,” said Sir Patrick Vallance as he drew things to a close, forgiving us for not fully understanding it.
You have to understand what the message is you want your audience to take away? You need to consider whether the language or jargon is appropriate? You have to make sure you have a robust understanding of the context and what you need to communicate. Are you sending the data over in a report or doing a presentation The latter may mean you can convey more in person rather than the page and vice versa for just a report. Tailor the design and complexity of the visualisations to match the target audience's level of familiarity with the data and their technical proficiency. If the visualisations will be shared digitally, ensure they are optimized for the specific platform or device, such as mobile responsiveness or interactive features.
Tip #4 - Focus attention with colour, shapes & font
A single chart can convey many points. Use colour to highlight the most prominent e.g. best performing product or indeed the worst performing product. When used sparingly, colour is one of the most powerful tools you have for capturing your audience's attention. For items important for context for the chart but not your main point, make them grey so they sit in the background. Use colour consistently throughout your report so it becomes easier for your audience to understand the data. So if you're comparing men and women and men is blue and women are pink (or indeed vice versa) then keep this colour coding the same throughout. Don't confuse your audience by changing colour! And you don't have to colour everything. If you just want to show the increase in the last 3 months on a line chart then just colour that and not the whole line. The audience's attention will immediately get drawn to the increase.
Use shapes to highlight areas of change, e.g. dotted lines for forecasts, boxes over data points you want to draw attention to. You can also use bold or italic fonts to draw attention to key points in quotes for qualitative research or data points which changes the picture.
Tip #5 - Tell your story with a focus on a call to action
Even the best visualisation may lack impact without a narrative. Every story has a beginning, a middle and an end - the beginning sets the scene and getting everyone on common ground, the middle is about convincing your audience that action is needed, the end is the call to action and tie it back to the beginning.
Here are some story building ideas...
Story idea 1 : Create a slide at the beginning of your point with bullet points of your story: 'Here's what we will cover in this report" - then organise your slides accordingly. It also sets your audience's expectations.
Story idea 2: Lead with the ending and start with the call to action and then provide evidence to support it.
Story idea 3: Be chronological - Take your audience through the same process your respondent did and this may build to a natural conclusion
However with individual visualisations, limit the number of take-aways to 1 ( 2 at most). I usually do this at the top of the slide, above the visualisation. Alternatively you can label the takeaway next to the key data point.
Before the end...
It is hard to succinctly provide tips on data visualisation and storytelling but hope the above was a little useful! 'Storytelling with data' is a book I'd highly recommend if you are trying to convey a data story. If you need help telling a story then get in touch: elliot.ferninsight.com