Innumeracy is the new illiteracy
In my recent articles, I noted that a significant challenge for many companies today is the vast amount of available data and their limited ability to use it effectively in decision-making. The core of this problem is mainly human-driven. Therefore, there’s a pressing need to build data literacy. If companies genuinely aspire to benefit from even a fraction of the data out there, they must elevate their overall competence with numbers.
88% of our potential audiences may struggle to read numbers, charts, or computations.
Based on a global literacy study, merely 12% of adults worldwide are numerically literate. In theory it means that 88% of our potential audiences may struggle to read numbers, charts, or computations.¹ That starkly contrasts the world’s literacy rate, which is currently over 86%.²
Remarkably, many societies overlook this issue. What’s more, numerous individuals take pride in their innumeracy.³ Have you ever heard excuses like:
- I’m not a numbers person.
- Statistics have never intrigued me.
- Physics seems like sorcery to me.
Or the "ultimate" one:
- I’m more of a humanist.
Is this truly a valid excuse?
Often, it isn’t a matter of lacking ability but instead not having the right tools. Consider this scenario:
We’re attempting to cross the street. We observe traffic approaching from both directions. We must judge if there’s sufficient time for us to cross safely. And that car in the distance? How quickly is it moving? Is its speed increasing? Decreasing? Has the driver noticed us?

In reality, crossing the street presents a complex probability challenge. Yet, we navigate it daily, often multiple times, and usually get it right.⁴
That’s just scratching the surface.
Think about driving a car at 60 mph. How many factors come into play? When driving, you’re also monitoring the speedometer and various other instruments. It’s not different to interpreting intricate visuals on a managerial dashboard. Then there’s the act of choosing an outfit in the morning. You’re making judgments and predictions that even sophisticated AI tools grapple with: forecasting the weather, determining what’s trendy, and matching colors.
The human brain processes around 74 GB of data daily.⁵ Remarkably, it still manages to take decisions. Meanwhile, my ChatGPT stumbled on a 200k-row Excel spreadsheet.

Using the "humanist" label isn’t a valid excuse, either. Consider dissecting a complex literary work, like a poem. Or grappling with a profound philosophical quandary? Or piecing together an intricate historical battle plan? Often, these tasks can be more challenging than constructing a statistical model.
Why do I mention all of this? My point is that innumeracy often doesn’t result from a lack of ability. Instead, it stems from the absence of tools and ineffective communication.
The irony is that many who claim innumeracy show proficiency in everyday tasks that demand intricate probability assessments and decision-making.
Why do we use data in communication?
Effective communication is pivotal in business; integrating data analytics can elevate it. Here’s how:
- Raise awareness: Use data to highlight issues, explain their causes, suggest solutions, and forecast results.
- Stressing importance: Data objectively prioritizes, guiding decisions like which projects to tackle first.
- Proving causation: Businesses often mix correlation with causation. Only the right data tools can clarify this.
- Storytelling foundation: Use data to craft a narrative, provide context, illuminate issues, and make conclusions.
- Predictive Power: Many companies rely on intuition for scenario planning. That’s a risky approach. Data offers a clearer path forward.⁶
Who should be responsible for talking about data?
After watching many YouTube videos and reading countless articles, the easy answer I could give is "data scientists." But that’s not correct. Allow me to elaborate on my reasoning.

Do you see the unicorn in the image above? Theoretically, it represents the "perfect data scientist" with an optimal blend of hard and soft skills. However, finding such individuals is as challenging as locating an actual unicorn. And when you do, make sure your pockets are deep enough.
Depending too much on such rare talents has its risks. Many businesses have faced setbacks when these top talents moved on to new opportunities. How can an organization mitigate this risk? It can cultivate these combined skills across a team rather than in one person. I’ll introduce a brief case study centered on a small team to illustrate my point better.
Mini case study: cooperation between data scientist and business analyst
Methodology
This mini case was prepared using ChatGPT Plus in conjunction with Advanced Data Analysis. Specifically, I utilized it to create synthetic data and produce data visualizations.
Setup
Envision a compact team comprising a data scientist (DS) and a business analyst (BA). Their mission is to probe and elucidate irregularities in daily sales for a retail company primarily driven by in-store purchases.
Cooperation Model
Step 1: DS & BA collaborate to refine the problem scope, determine the data to analyze, establish the time frame, and select the most effective communication method.
Step 2: DS then embarks on the analysis, proceeding as follows:

Step 3. DS employs a method combining moving averages with standard deviation. This facilitates the detection of anomalies in the sales figures. A data point falling outside a set threshold (2 standard deviations away from the moving average) is flagged as an anomaly.
Step 4. Upon completion, DS conveys the findings to the BA.

Step 5. BA deciphers the anomalies:
- Promotion at our store (days 50–60, end of February): An uptick here is likely driven by an in-store promotion, leading to heightened customer visits and purchases.
- Competitor’s new store launch (days 150–160, early June): A new store nearby might be taking some of our customers, causing our sales to drop.
- Temporary store closure (days 250–260, mid-September): A pronounced sales drop in September suggests one of our outlets temporarily ceased operations, possibly for refurbishments or unexpected events.
- Another promotion (days 320–330, late November): A surge in sales during this period probably stems from another promotional or sales campaign in our store.
Step 6. BA and DS review the findings together. This collaboration provides DS with insights into the business side while BA enhances his or her data proficiency.
Step 7. Complemented by relevant recommendations, the concluded analysis is presented to the concerned stakeholders.
Conclusion
This concise case study underscores the synergistic relationship between a data scientist and a business analyst. While the data scientist focuses on anomaly detection through statistical techniques, the business analyst contextualizes these findings, depicting the underlying causes and delivering practical insights. As illustrated, robust communication between both roles is central to the success of this collaboration.
Just as those who are "innumerate" might shy away from numbers and graphs, the "numerate" often hesitate to engage in human interactions.
Communication, too, is a science
Just as those who are "innumerate" might shy away from numbers and graphs, the "numerate" often hesitate to engage in human interactions.
You know what? That’s a profoundly unjust stereotype.
While I am acquainted with several numerate individuals who communicate effortlessly, even if you’re an introverted numerate, you can harness your skills and approach communication as methodically as any statistical challenge.
Communication can be distilled into three primary components:

This simple model is not only an approach to structure communication. It also offers me an instant structure for my presentations, which I can consistently apply.
Now, let me provide some firsthand guidance on addressing these three components.
WHY: Mind the others
Context is king, as evidenced in one of my latest articles. It contributes to a staggering 80% of a narrative’s impact. The backdrop of a story spans three core facets: situational, functional, and data-driven. This foundational information equips the narrator and listeners, addressing pivotal inquiries about the tale’s audience, central theme, and delivery style. Embracing and weaving in context is vital for the story to connect and be comprehended by its target listeners.
A common oversight among data professionals is leaning heavily on technical language. This can easily occur when we neglect the context. For instance, is it apt to use certain terms for a specific audience? Phrases not commonly used outside of academia, such as "cohort" or "longitudinal," and terms with varied interpretations, like "surveillance," can be confusing. Similarly, concepts like "probability" or "risk" could be misunderstood. If these terms are indispensable, aim to present them comprehensibly. Avoid getting lost in the weeds of detailed arguments, even if they’re meticulously crafted. It’s beneficial to understand that many operate based on intuitive shortcuts rather than logical models, influenced by biases like anchoring, misunderstanding cause and effect, disregarding randomness, and many others.⁶
HOW: Craft good stories
The impact of well-crafted stories in corporate communication is clear, mainly when supported by data. But for this discussion, I’d like to delve into how these stories are constructed. Consider the narrative arc as championed by Brent Dykes.

Long beginnings can spoil stories that use data. This can happen on purpose or by accident. For instance, the audience might suddenly shorten a well-organized story, causing us to make cuts in it. Sometimes, it’s just because the story was badly planned. How can we avoid these problems?
Cole Nussbaumer Knafflic, the renowned author of a global bestseller⁷, recommends one approach: 3-Minute Story.
Consider crafting a concise, 3-minute rendition of your presentation. This approach offers dual advantages. Firstly, it serves as a contingency if your allotted time gets unexpectedly reduced from 20 minutes to a mere 3–5. Secondly, it acts as a refining tool. If you can streamline your presentation effectively and quickly down to 3 minutes, trimming excess from the full version becomes effortless.
Rehearsal
Another tactic I advocate is rehearsal. Whenever possible, practice your presentation, ideally aloud. While rehearsing in front of a mirror is beneficial, having a trusted colleague provide feedback can be even more valuable.
HOW: Use analogies
An analogy draws a parallel between two things, often to elucidate or clarify. Take, for instance, the "crossing the street" example I used earlier. Employing an analogy, we aim to demystify complex analyses by presenting them through more familiar scenarios.

The concept behind an analogy is to shift the perspective from the specific data project to a broader, more familiar context, explain it there, and then bring the understanding back to the original data context. Analogies act as helpful connectors, linking complex data subjects to everyday situations, making them easier to grasp. To craft a meaningful analogy, it’s vital to deeply understand the data, brainstorm relatable scenarios, assess their fit, and then refocus on the data’s core issue.
So, where can one find good analogies? Real-life examples often hit the mark. However, the challenge is having them on hand when needed. A modern solution is to consult ChatGPT for creative suggestions. Below, I’ve provided analogy examples related to the mini-case study discussed earlier.

Another strategy requires more dedication: consistently gathering and storing analogy ideas. This could be in a physical notebook, a digital tablet, or any method that suits you. The crucial part is keeping these notes structured and within reach.
Yet, there’s a cautionary note regarding analogies. There’s the danger of overextending them. The more unfamiliar you are with the context, the likelier you might stray. If people think the comparison is too simple, rude or childish, it could be a problem.

HOW: Mind the timing
To manage the problem of timing, I’ve set a guideline for myself. Typically, I allocate roughly 1.5 minutes for intricate, data-driven presentations for each slide, encompassing the title, agenda, break slides, etc. I don’t differentiate between slides packed with content and those more concise. Over a more extended presentation, this tends to balance out. For less complex talks where slides aren’t as data-intensive, I might reduce this to 1 minute per slide. In such case I always prioritize rehearsing these presentations in advance. Sometimes, this reduction is an estimate based on prior rehearsing.
If I’m given a specific time, like 15 minutes, especially for presentations to senior executives, I determine the maximum number of slides using the following formula:

Nslides
stands for maximum number of slides in the presentation (rounded to the nearest integer value). Tminutes
is the allotted time. For brief presentations with fewer than ten slides, I forgo including agenda or Q&A slides.
For a 15-minute slot, this equation allows me to produce a maximum of 7 slides. Hence, I’d structure my presentation with a title slide, followed by a maximum of 6 content slides. This would break down to one for context, up to four for analysis, and one dedicated to primary conclusions and a call to action (CTA). If necessary, I might slightly adjust this distribution, perhaps saving a distinct slide for the CTA.
Why such brevity? It’s usually sufficient. Plus, we need to account for unforeseen circumstances, such as:
- Extended time connecting to display equipment
- Key attendees arriving late
- Allotting time for active discussions
- Any other unforeseeable interruptions or delays.
HOW: Mind the numbers
A crucial aspect to ponder is our approach to presenting numbers. To elucidate, I’ll provide an example. First, let’s examine the "incorrect" chart.

And now the corrected one.

See the distinction?
The second chart showcases the "meaningful digits" guideline. Three seem to be the optimal number of digits to display from my encounters. Using only two can be tempting, but for vital figures like company revenue, there’s a risk of over-simplification, which may mislead.
Then there’s the "friendly numbers" principle. It’s about presenting figures in the most straightforward, memorable manner. For instance, rather than detailing USD 2,954,321.51, we might round to roughly USD 3 million. But use this method with caution: it’s a slippery slope to unintentionally skewing data, potentially misleading your audience.
HOW: Embrace the power of titles
Titles play a pivotal role in data-driven storytelling. A compelling title captivates the audience and often provides essential insights, negating the need for detailed data analysis from tables or charts. It empowers them to decide whether to delve into the content or await more relevant information.
Consider leveraging the "power words" technique from journalism and the Internet when crafting titles. These words enable you to manage attention and set expectations deftly. Phrases like "a quick win," "the only viable solution," or "a red flag in the project" can be potent. However, exercise caution with power words: if we initially present a solution as simple and swift but the ensuing content reveals a complex challenge, our credibility may be compromised.
Additionally, it’s essential to avoid purely technical titles, such as "Number of incidents reported to IT." Instead, opt for functional titles that provide context, like "Immediately after implementation, the number of incidents reported increased." This ensures that your audience can readily grasp the relevance and implications of the content.
HOW: Use color strategically
Color is a powerful tool in data-driven communication. However, it’s a tool that demands responsible usage. We should resist the temptation to employ color purely for decorative purposes. In all instances, color should carry functional significance, highlighting essential information. To illustrate, let’s consider the charts below.
Decorative use of color:

Strategic (functional) use of color:

HOW: Choose visuals wisely
As explored in this article, three visuals are reliable and can convey a wide range of data-related messages without confusion. While this list isn’t exhaustive, adhering to it significantly enhances the likelihood of success.

Using These Three Visuals:
- Column (or bar) chart: Ideal for comparing cumulative or discrete values. Can be used for trend presentation, provided there are no major shifts.
- Line chart: Effective for showcasing directions and changes in trends, especially for continuous variables. Multiple lines can be compared on a single chart.
- Pie chart: Suitable for displaying the contribution of a part to the whole (e.g., the percentage of revenue generated by a specific product compared to total sales). Avoid using pie charts for comparisons at all costs!

When creating charts, adhere to the mentioned rules (e.g., color and titles) and consider proper scaling (particularly the vertical axis, which should typically start from 0 unless there’s a valid reason not to). Simplify charts by removing extraneous elements like frames, gridlines, axis labels, or legends.
WHAT: use CTA
This article’s final piece of advice concerns using a Call to Action (CTA).
Without a CTA, our presentation remains, at best, an intriguing or perhaps even appealing piece of information. A well-crafted CTA imparts a distinct character to data-driven storytelling.
What renders a CTA effective?
- Utilizing active rather than passive voice.
- Employing imperatives with care.
- Clearly identifying individuals or teams responsible for executing agreed-upon actions.
- Promptly following up on decisions immediately after the presentation and, subsequently, at short intervals to ensure progress.
Conclusions

In data communication, there’s a significant gap between numeracy and innumeracy. While just 12% of adults globally are numerically literate, the majority struggle with data comprehension. Yet, the irony is that many who claim innumeracy show proficiency in everyday tasks that demand intricate probability assessments and decision-making. The challenge lies not in their inherent ability but in the tools and methods used to convey data.
Effective communication strategies bridge this divide by crafting relatable analogies, strategically using visuals, and employing thoughtful storytelling. By grasping the audience’s context and reducing technical jargon, data professionals can enhance accessibility.
Ultimately, data storytelling and communication revolve around simplifying complexity, connecting with the audience, and driving actionable insights. Whether you’re a data scientist or a business analyst, the key is to converse in your audience’s language, ensuring that the power of data remains clear.
Did you like this article? Consider a subscription to get notified about new stories I publish or follow me.
Resources
- Ann Wylie: What are numeracy rates by country?
- World Population Review: Literacy Rate by Country 2023
- Missouri University of Science and Technology: Innumeracy: The Product of Misrepresentation
- Rebecca Nugent | TEDxCMU: We’re All Data Scientists
- Wonder: How Much Information Does the Human Brain Learn Every Day?
- National Cancer Institute: Making Data Talk: A Workbook
- Cole Nussbaumer Knafflic: Storytelling with Data: A Data Visualization Guide for Business Professionals