Data Storytelling is not Data Analysis
with apologies, or at least indebtedness, to Stuart Ritchie
Data storytelling is a powerful way to turn often complex data into an interesting and engaging story. It involves analysing and interpreting data to uncover patterns and insights, then using those insights to create a compelling narrative that conveys a meaningful message. Storytelling can be used to explain complex topics, make decisions, or motivate and inspire action. It can also be used to communicate emotions and feelings, creating stories that are both informative and captivating.
Sounds good, doesn’t it? It’s from an article I wrote a few years back. But there’s more, which I didn’t think about enough at the time, but certainly do now. Data storytelling too often is caught up in the story and not grounded in the data.
Data Storytelling can be used to distort our thinking, to evade difficult truths in the data and to simply deceive. In the worst scenarios - from my point of view - this happens without the storyteller even noticing their own misrepresentations.
Science is not Storytelling
In a recent blog, Stuart Ritchie was, as ever, admirably clear that Science isn’t Storytelling. Stuart is the author of an excellent book, Science Fictions: The Epidemic of Fraud, Bias, Negligence and Hype in Science, which lays bare many of the distortions of modern scientific practice. In his blog he took to task a recommendation that scientists should write papers with their conclusions in mind, paying attention to narrative, framing and (save us!) take-home messages.
You might find Stuart’s approach to science rather austere …
A scientific paper isn’t supposed to be a story. It’s not an opportunity for you to flex your creative muscles, or to craft a tale ... If it’s boring or doesn’t have a “narrative” - well, that’s sometimes just how it is, because in reality scientific studies often career off in unexpected directions that belie attempts to condense them into a clear, linear, account.
Of course, as an engaging and lively science writer himself, Stuart knows the value of good writing. But his concern here is absolutely valid: focussing on the story can distort our analysis.
And so to Data Storytelling …
Data storytelling is not a branch of marketing
The first and most significant problem I have with the narrative approach to data is that too often, the story is all there is. In business analytics we very rarely have the equivalent of a scientific paper. All too often I see people jump straight from analysing to storytelling. Sometimes there’s not even a jump. Telling the story becomes the analytic method. Popular business intelligence tools reinforce this thinking. The story becomes our primary artefact often crafted by the user directly in the same tools - at the same time - as they are analysing. As Stuart says:
the writing of the paper is where much of the bad stuff comes in. For so many scientific projects, the “doing the research” part and the “writing up the research” part are simultaneous.
There are, of course, many differences between business intelligence and scientific research. For one thing, business analysis tends to involve the ongoing monitoring of changing data looking for trends and emerging features, then when something is found reporting on it in order to direct decisions based on the findings. So there’s a reasonable argument to made that data storytelling has a role to play here, because there is a motivating purpose - to persuade people to take action. Scientific research is a project to discover what is true, not a manifesto for action.
But there’s a danger in business taking this shortcut from analysis to action. Our business thinking is already deeply motivated, and rightly so, by profit, promotion, and productivity. The biases these motivations introduce could be addressed more effectively if we separated out analysis from storytelling.
Find the truth first, then tell the story.
Unfortunately the hype around data storytelling and the affordances designed into the analytic user experience make this harder path less likely to be chosen.
Profound point that is seldom mentioned. Do not tell a story about your data analysis for which the ending is predefined (as it usually is for normal stories). Thanks...
However, if data storytelling is not data analysis, then how should you tell the story behind your data analysis? Topic of your next blog? Is it like a story told by Walter Isaacson about Steve Jobs or Da Vinci? ...true to facts but profound in insights? But then, where is the call-to-action? Or, should there be?
In our AI ("Detect & Alert" insights - designed by Aldecis), Data StoryTelling is used to give explanations of why the AI highlighted these particular insights.
Because to avoid the "BlackBox" nightmare of AI, nothing (nor several graphs, nor several tables) is better than an explanation in Natural Language.
Not only the Data StoryTelling provides "explanability", but our Chatbot also help users to answer the next 3 questions
1) Helping the user to find the Root cause (There is no cure without diagnosing the problem first)
2) checking that his action plans are solid enough (to be submited to the panel of experts that decides between several proposals)
3) proposing who should be included in this collaborative decision process (Collective Intelligence)