What does it mean to become data-literate?
According to Brent Dykes , data literacy has two distinct facets: data analysis and data interpretation. While they’re often perceived as a single process, distinguishing between them is crucial. Data analysis is technical, encompassing tasks from data cleaning and anomaly detection to trend analysis and identifying correlations. On the other hand, data interpretation delves into attributing meaning to these findings. It asks questions like:
What’s driving the observed trend? How can our organization capitalize on a specific correlation? Is a detected anomaly a rare occurrence, an unseen opportunity, or perhaps, in some cases, does the observed correlation indicate causation?
By the way: obtaining a ‘Yes’ answer to the last of the above questions is often considered the holy grail of business analysis.
Not so long ago, I looked at the latest ‘BARC Data Culture Survey.’ It said that in 2022, only 32% of participants made decisions purely based on data. 18% used their experience or just went with their gut. The other 51% used a mix of both . This made me think: Is this last number a bad thing?
On the one hand, it is not. Despite the advancement in data science, including the deployment of LLMs, this ‘human touch’ is still needed. Only humans can understand humans: their moods, tastes, and emotions. No matter how complex, a computer program still remains inferior to us in that field. On the other hand, overuse of human elements might be sometimes harmful. It all depends on how much of each ingredient (i.e., data or gut) is used and for what kind of decision they are used. I’ve seen that the bigger the decision, the more people go with what they believe or have seen before. And that’s perhaps not the right direction nowadays.
Why it makes a lot of sense to separate data analysis from data interpretation?
Well, that’s another ‘human thing.’ One of the biggest challenges is switching between detailed and ‘helicopter’ views (and the other way around). You should not expect a manager, already overloaded with many responsibilities, to play with a given dataset for a couple of hours or days. On the other hand, data scientists may lack the necessary background knowledge, such as shifts in strategy, understanding of the company’s environment, and critical decisions taken and not yet announced.
Given their myriad responsibilities, managers often rely on data professionals to conduct detailed analyses. These specialists, after extensive data exploration, offer invaluable insights. Yet, leaders must engage in data interpretation rather than passively accept conclusions.
Involvement of managers ensures that data interpretations align with the broader organizational context.
When leaders build on the foundational insights from data professionals, the collective reasoning results in more comprehensive decisions.
The aim of improving the data literacy level in the organization isn’t to craft data scientists out of every employee but to nurture data-savvy individuals. The emphasis is on equipping most of your team with basic data skills for their regular tasks rather than intensive data-driven roles.
Being data literate means having the confidence to manage everyday data without advanced statistics or programming knowledge.
This foundational proficiency unlocks your organization’s data potential. To determine what constitutes data literacy for your team, pinpoint the ’Minimum Viable Proficiency’ across the data hierarchy tiers rather than listing all possible skills. Brent Dykes suggests that we could break down the required skills into a 3×3 matrix, where we put levels of data literacy (vertical axis) and necessary (minimum) crafts for each level (horizontal axis) .
In the Read category, the Data level involves basic numeracy and domain-specific metric understanding. The Information level adds the ability to understand visuals (graphs) and a basic statistical grasp, while the Communicate stage emphasizes data interpretation and critical assessment.
For Work With, the Data level prioritizes familiarity with analytics tools and data manipulation. The Information level focuses on descriptive and diagnostic analysis skills, and the Insight level underscores decision-making based on data insights.
In the Communicate sphere, Data level skills include addressing requests and discussing data topics. The Information level highlights data presentations and visualizations, with the Insight stage emphasizing the art of impactful data storytelling .
Following the 80/20 principle, I believe the Insight level is essential to attain data literacy. While other levels are beneficial, they aren’t critical, especially from the decision-makers viewpoint.
First and foremost, it is imperative for everyone within the organization — especially managers and leaders — to cultivate a foundational understanding of data science tools and techniques. This does not suggest that everyone should embark on statistics courses or dive into learning Python or R. Rather, the aim is to grasp essential terminology, understand the advantages and drawbacks of specific tools, and be equipped to conduct rudimentary analyses. While I recognize the time constraints that often exist in corporate roles, I’ve found that even a basic comprehension elevates my interactions with data scientists, eliminating the need to revisit foundational concepts and improve collaboration. I tested this on myself.
‘Work with’ sphere
Changing how we work with data could significantly shift our company culture. Strong leadership is vital. If leaders aren’t comfortable with data, they’ll trust their gut over the numbers.
I believe we can make use of ‘data champions’ in our teams. These are people passionate about data and willing to help others understand it. They show that data and analysis are not just about complex tables but also about exciting discoveries.
We also need user-friendly tools that let everyone explore and understand data even without tech skills. Managers should use effective, well-designed dashboards to get a clear picture quickly .
However, there are risks. With all the hype around valuable data, some companies collect too much, wasting time analyzing it and slowing decisions.
Also, we might lose enthusiasm if we don’t see quick benefits from using data. Celebrating small successes is vital, especially if we’re new to a data-driven approach.
Here is where ‘storytelling with data’ comes into play. Why is it essential? Storytelling, through narrative and visual aids, helps make sense of vast data. Compelling storytelling can bring clarity to any story, no matter how complicated. This skill is so vital that Forbes lists it as a top requirement for data scientists. Yet, many analysts struggle with it. Effective data stories:
- Make complex topics relatable and shareable.
- Drive decision changes more than personal experiences.
- Simplify results from intricate data analytics.
Compelling storytelling refers to the “so what” statement: the core message. Many analysts lack design creativity or fear simplification. Yet, unclear visuals can hinder business decisions .
I plan to write about my approach to storytelling with data soon: stay tuned!