This is the second in a series of posts by Kirk Borne, Chief Science Officer at DataPrime, an AI- and data-driven jobs platform for the data sector. Kirk is also a consultant, astrophysicist, data scientist, blogger, data literacy advocate and renowned speaker, and is one of the most recognized names in the industry.
In my first article for this website, I discussed steps that organizations can take to build data literacy and data fluency in their workforce. Across the spectrum of roles that require data fluency (data managers, data analysts, data scientists, data engineers, machine learning and AI specialists, and data officers), one deserves special attention: the data storyteller.
But first, I have a short story that’s relevant to data storytelling.
Back when I was in middle school, one of my homework assignments was to write a four-page essay related to a historical event. I couldn’t think of four pages of original things to say about the topic, so I essentially repeated myself several times, saying the same thing using a variety of different colorful adjectives and adverbs. I expected to receive a bad grade on the assignment, but the teacher had mercy on me and gave me a decent grade. Two essential facts remained: (a) I didn’t really know much about the topic I was discussing; and (b) I got away with it because I had a sympathetic “client”.
Before I prove the relevance of this story to our topic, let’s bring the focus back to the data storyteller with three declarations. First, data storytelling is a desirable characteristic of all other data-fluent roles. Second, data storytelling helps create business value from data in a broader context than some of the other job-specific requirements in the different data-fluent roles. Third, because of these first two statements, the data storyteller may not be a stand-alone role in organizations, though it could be, particularly for data journalists.
In this blog I will validate those three statements, by presenting my list of required aptitudes of data-fluent individuals that apply generally across the full spectrum of roles. Then, I will present a taxonomy of data storytelling abilities that will necessarily differ across those roles.
Aptitudes of data fluent individuals
Aptitudes are not skills, not only because the former are general and the latter are specific, but primarily because skills can be taught (and learned) whereas aptitudes generally come with the individual. For example, not every person has the aptitude to be a top chef. People can be taught to cook, but there is more to to being a top chef than cooking skills.
The aptitudes of successful data-fluent professionals are held by individuals who ‘sail the seven seas of data fluency aptitudes’.
Why seven seas? Mostly because it’s a nice metaphor for exploring vast and endless seas of data. But also because all the aptitudes I’m about to list start with ‘C’ (though in fact, there are more than seven). These are: critical thinker, curious, creative, collaborative, computational, courageous problem-solver, cool under pressure, community-focused, continuous life-long learner, consultative, compassionate and communicator.
The last three aptitudes are especially important for data storytelling:
- Consultative – being able to listen to a client’s needs, repeat them back successfully, and elicit specific requirements for what the client wants you to deliver
- Compassionate – having empathy with your clients and audiences who may not understand all the jargon, mathematical complexities, emerging technologies, and coding snippets laid before them
- Being a good communicator (obviously)
You may have a sympathetic audience (like my school assignment), but we should not assume it will work in our favor (like it did for my 13-year self).
Different storytelling requirements
Not every data-fluent role has the same data storytelling requirements. I have developed a broad three-tier taxonomy of such abilities:
- Talk the talk
Can use the essential words (including jargon) in a sentence, but without conveying much nuance or depth of meaning to the story. This approach may work in a marketing or sales pitch, or with a sympathetic client and be sufficient for those staff in your organization.
- Walk the talk
Can do the hard stuff (the data science, data engineering, cloud DevOps, AI/ML modelling, or Python coding), though can’t necessarily explain those hard things to a general non-expert client or audience. This will work well with a tech-savvy client, which may be acceptable for data scientists, AI/ML specialists and data engineers, but may turn off a non-technical client.
- Talk the walk
Can communicate hard concepts, difficult engineering or science methods, and complex technologies in terms that non-experts can comprehend; incorporate into their decision-making processes; remember; and actively engage in conversation. Talking the walk requires you to bring your technical results back up to the level of communicating and summarizing the strategic or tactical value to stakeholders. This is essential for data analysts, data managers, data officers, data thought leaders, data journalists, and (of course) the data storytellers. Talking the walk is always a winning approach to data storytelling with all stakeholders.
Before I close with a data story, I present four quotes that underscore the power of storytelling, both in a data context and in a human context:
“People will forget what you said and what you did, but they will never forget how you made them feel.”
“No one ever made a decision because of a number. They need a story.”
Daniel Kahneman and Amos Tversky.
“Statistics use the analytical part of the brain whereas stories engage both the analytical part of the brain as well as the emotional part of the brain.”
Ellen Lingwall on the neuroscience behind data storytelling.
“Storytelling is the oldest form of teaching.”
An age-old adage.
Much as your corporate communications strategy should reach, inform, and inspire the broadest community of your business stakeholders (employees, management, customers, investors, market analysts, and even social influencers), so should your data storytelling, which can motivate positive value-producing actions by and for all those groups. Data storytelling may actually be a more valuable business workforce asset than all the data technology that supports it.
A true data story
I end with an example from FINSERV, an anonymized financial services firm managing more than one trillion dollars in long-term investment assets for customers. FINSERV sought to identify an actionable early warning signal in its customer transaction data to identify customers at risk of closing their account and moving their investments elsewhere.
FINSERV managers asked themselves: “Can we use data analytics to predict the churn in advance, and then take some customer care outreach action to prevent it?” They suspected that a precursor signal of customer churn might be something as simple as an increased frequency of customer visits to their online investment account. Such a sudden increase would be unusual for long-term retirement accounts. FINSERV hypothesized that an increased frequency might mean that the customer is comparison-shopping with another investment firm.
Consequently, FINSERV’s board authorized a $1m investment in an analytics experiment: a standard A/B test, which tested various thresholds for the number of online account accesses to determine the optimum point at which FINSERV would reach out to the customer with some “extra special” customer care. FINSERV would send those customers some “gentle” reminders about their investment options and new investment opportunities, plus pointers to their investment performance dashboard, some new investment calculators, and the updated investment FAQs. Some customers above the threshold of account accesses in the month received this customer care treatment (the “A” treatment group), and some customers did not (the “B” control group).
After one quarter the analytics team reported back to the board on the outcome from their $1m analytic experiment. Not only had the analytic strategy significantly reduced customer churn, thereby improving retention, the analytic actions had also saved the business over one billion dollars in customer value (a 100,000 percent ROI).
A simple algorithm (crossing a threshold) on some simple data (web clicks) had a profound outcome for the business, which could be simply explained and understood by everyone involved.