Storytelling Through Data
Data visualisation can help you see what could otherwise have been missed, should you have looked only at a table of figures. Good visualisations help tell a well-rounded story about your data.
Trying to find patterns around us is an instinct as humans, and the same is true when analysing data. Good visualisations come from good analysis, highlighting trends, patterns and relationships in your data. Your presentation should highlight their relative importance to the story you are telling.
Trends, Patterns and Relationships
A relationship in data is a general term for correlation or dependence between variables, whether casual or not. For example, consider a scenario where as variable X increases, variable Y decreases. Is this a coincidence? Is one responsible for the other? Are there external factors causing this relationship?
A trend is the general direction or change of a variable, over time (or its relevant metric). A pattern is a set of data that follows a recognisable form, such as data trending in a certain way, or the general tendency of data to move in a certain direction.
Looking for trends that emerge over time, across space or between categories allows us to find patterns between data-points, which can be further explored to determine a relationship between variables.
The way we chose to display our data, or the story we tell, will affect how engaged our audience is, and how much of the information they retain. Below is a simple, yet effective, rubric for telling the story of technical data.
The Three-Act Structure
When in doubt, look to Aristotle.
Every tragedy must have a beginning, a middle, and an end. The same is true for every story you’ll tell through data. Aristotle’s insistence on this idea led to the three-act structure we are used to in film and literature, and this concept lends well to presenting data-heavy ideas.
Act One, the Setup
This is where you provide context about yourself, the entity the data represents, and it may interest your audience to know how and why you chose your research questions. The aim of Act One is to situate yourself in both the business-case of your data, as well as in the technical aspects you’ll cover.
Act Two, the Confrontation
This is where the bulk of your analysis will be shared, but remember to build up the most impactful observations, so your audience is clear on which information is the most important take-away. Act Two is where you will make the most use of visual aids, and where most of the technical information will be shared. Be careful not to fact-dump, though, as we need the audience to stay engaged!
Act Three, the Resolution
This is where you relate your findings back to your research questions, and ground your results in the context you have already set up. Act Three will reinforce the importance of your findings, commenting on where this information sits in a wider context, and share successes and limitations of the work you have done. If relevant, this is where you indicate how your research aids business goals, or make suggestions on what could be done to turn these insights into quantifiable change.
To communicate these ideas to your audience without visualisations could be tricky, especially expressing the significance between numbers that only differ by a small fraction. Here are some things to consider when creating visual aids, to maximise their efficiency, and minimise confusion.
1. Keep It Simple, Keep It Specific
Bite-sized chunks of data will always be easier to digest.
Often, you will be displaying high-level, overview information. So, show just that: an overview. A common mistake is to get too in-depth too quickly, which may confuse your audience, or lose their attention. Instead, keep the information simple and focus on one talking-point per visual aid.
Another way to lose your audience is to focus on information outside of their subject of interest. As mentioned, some context is needed to ground your data within your story, and may inform some of the choices you have made. However, focusing too much on the back-story can slow the pace of your story, and can overload your audience with information, which can then cause confusion as you dive deeper into your analysis.
Keep it as simple as possible, and remember to use language and ideas that are suited to your audience.
2. Stay in Scale
May the scales tip ever in your favour!
The scale of your visualisation will directly impact how important your audience believes a trend, pattern or relationship is. Make sure you aren’t misleading them!
For example, take the charts above. Both are displaying the same information, but with a different scale on the Y-Axis. Either graph could be an appropriate way to display this information, depending on the message we were trying to get across.
If a difference of £1500 is significant in a particular context, it makes sense for your axis to show this significance, as in the visual on the right. However, if this difference only has a minor significance, it may be more sensible to model similar to the left visual. It will be for you, the expert, to determine how significant these changes are, and how to represent this significance.
3. Visually appealing
Can your graph walk AND chew gum?
It’s no secret, that things that look nice, keep us better engaged. It’s in our nature to scan through a picture or document, searching for things that stand out, so use this to your advantage!
Leave as much whitespace as possible, and using sizing effectively, so the visual hierarchy is clear to your audience. Visual cues are helpful to display ideas without having to use words: think ‘danger’ being represented by a skull and cross-bones, or ‘money’ being represented as a dollar sign. Using these (where possible) can allow for effective communication of concepts, without adding too much clutter to your graph. Annotate where necessary, but not at the expense of a clean, clear visual.
Revisiting the importance of relevance, having relevant visual aids can make a huge difference in ease of understanding. If your data is representative of difference places, why not represent the data on a map? Heat-maps can be particularly effective in this case.
Animated graphs can do wonders for engagement, as well as information display, but if they are too busy the message may be lost. Keep the transitions slow, and only use one metric at a time.
As data-driven decision making becomes the norm across domains, the importance of telling data-heavy stories in on the increase. A captivating story told through insightful visualisations can be invaluable. Let me know what you think makes a great visualisation!
