The world’s leading publication for data science, AI, and ML professionals.

From Numbers to Actions: Making Data Work for Companies

What flops and what works?

Today, organizations and individuals are swamped with data. Each day, 329 million terabytes of data is produced globally, amassing a whooping total of 120 zettabytes per annum [1].

But what does that mean in tangible terms? Consider an iPhone with 128 gigabytes of storage. Let us imagine that every day, the equivalent of three billion iPhones is packed with data*. Sounds impressive? Let’s delve deeper. This daily figure extrapolates to a theoretical 1.08 trillion iPhones overloaded with data annually. Given the current global population of approximately 7.8 billion, this means each person would need to possess nearly 139 iPhones [2]. **** Absurd, right?

The sheer volume is staggering, but the growth rate is equally astonishing. Just 13 years ago, in 2010, the annual data creation stood at a comparatively modest two zettabytes…

That’s just scratching the surface. Consider the data that never make it online – files stored directly on our devices or notes and documents penned on paper (yes, that’s still a thing!). Estimating that volume?

I wouldn’t even venture a guess.

So, there’s a ton of data out there.

But what does this mean for businesses?


I recently reviewed the latest Data and AI Leadership Executive Survey [3]. The results showed that 97% of companies have already invested in data and related infrastructure. 92% have put money into big data and Artificial Intelligence. You’d think this means they’re seeing returns on these investments. Not quite. A mere 40% of those surveyed said they view data as a ‘revenue-generating’ asset.

Only 27% of these companies consider themselves data-driven organizations.

What? Why?

The main issues are twofold. While some problems stem from technology, a massive 92% arise from human factors like organizational culture, people, and processes. Even tech issues often boil down to human errors.

This might surprise some. I once saw a statement that mistakenly said "decision-driven data" instead of "data-driven decisions." A simple typo? Perhaps. Freud might have had a theory or two on this, and I’d happily chime in. While it seemed like a blunder, my research indicated otherwise. The deeper I dug, the more justifications I found for this curious juxtaposition.

It’s impossible to cover every nuance in a brief article, but I’ve highlighted a few points to illustrate the complexity for you, dear readers.

Data swamp

Starting with a common myth: many believe the issue lies within the data, analysis, or tools. Not the people.

Over time, data supply mechanisms have evolved. The first-generation system, Data Warehouse, moved data from operational systems to business intelligence platforms, simplifying queries but suffering from complex processes and handling diverse data. The second-generation, Data Lake, kept data in its original form, benefiting machine learning but introduced its own challenges like data quality issues and source mapping problems. The third-generation, Cloud-based Data Lake, merges features of its predecessors in the cloud, offering real-time data access but remains complex and can be slow in converting data to insights. The fourth generation, Data Mesh is a decentralized approach to data architecture and organizational structure, promoting domain-oriented ownership of distributed data products [4].

Too often, however, companies start with a data warehouse, data lake or data mesh, yet they end up in the same place: a data swamp. Think of it as a messy closet of data.

Without clear guidelines, data can become disorganized and chaotic, similar to a game without rules or a closet filled with unlabeled boxes. Companies often store unchecked data, making its usefulness and accuracy questionable. Uploading data without a strategy further adds to the confusion, much like carelessly throwing clothes into a closet. And, if not maintained regularly, a data lake, like an ignored garden, can deteriorate into a swamp.

In short, to have a useful data lake, you need to organize it well, just like keeping a room clean. If not, it’s easy to end up with…

Where is my insight?

Let me share an experience from my corporate life. During a routine monthly financial review, I spotted a significant discrepancy in one of the cost lines, let’s say it was a million US dollars. When I inquired with an analyst about the discrepancy, this was the breakdown I received:

  • 100k US dollars: specific reason
  • 50k US dollars: another reason
  • The remaining amount? Simply labeled as ‘various’.

My face must’ve turned a shade of pale seeing that, especially since I had an imminent meeting with the management board to discuss these variances. I was caught off guard with nothing substantial to present.

This highlights a frequent issue with data insights: they tend to skim the surface, often only stating what’s already evident. We could easily use tools like ChatGPT for such surface-level information. However, despite the power of advanced models like LLMs, we still expect humans to deeply analyze and interpret data. We want them to derive meaningful conclusions that can be turned into actionable steps. The insights should offer value and suggest specific actions, not just relay information that one could easily find on their own.

The fish rots from the head down

Brent Dykes, in a LinkedIn post, quotes the story of Toyotomi Hideyoshi and his failed Korean invasion from 1592-1598.

This leader was notorious for his ruthless nature. Bringing him bad news could end up in the death of the messenger. Due to this, his generals hesitated to inform him of the true status of the campaign, which was far from successful. They played their cards so well that Hideyoshi remained under the illusion of success until his last breath. Once he was gone, Japan swiftly retreated from this conquest.

This story highlights the clash between objective insights backed by solid data and certain management approaches. In this case, one is driven by fear. It’s not an only possible scenario. Consider presenting an insight during a meeting that makes a few attendees uneasy. Even if the information is rooted in hard facts, the majority of individuals (from my observation) would defend their stand. This underscores the importance of grounding our analysis in objective and verified data. Moreover, if there’s potential for discomfort, it’s wise to align with those who might feel targeted, ensuring a more receptive audience.

How difficult could be to prove the obvious?

First, let us watch a video interview with the famous Sir Rowan Atkinson.

Sir Rowan Atkinson is indeed a comedic legend. It’s amusing to think that someone might not recognize him as the talent behind the iconic Mr. Bean. Yet, sometimes, the more we try to convince others of an evident truth, the less believable it becomes.

Many of us have been in similar situations. I remember personal experience trying to elucidate the intricacies of accrual-based accounting to sales reps. It was a very "special" moment. Sometimes, the gap between expertise and understanding feels vast, and no amount of explanation can bridge it. It’s a reminder that communication isn’t just about what we say, but how it’s received and understood by our audience.

Oh, the joys and challenges of communicating complex topics! For those in the Data Science community, it’s a familiar scenario. You dive deep into analysis, employing advanced techniques that you find utterly fascinating. Yet, when you present your findings, you’re met with blank stares or polite nods. It’s as if you were speaking a foreign language.

Did they genuinely get it? Or were they just nodding along, trying not to look lost? Often, the absence of questions is a telling sign. Unless, of course, you’ve primed someone in the audience to throw you a softball query. That is a good idea by the way.

This highlights the critical importance of not just mastering the technical aspects of one’s field but also being adept at translating that knowledge into relatable insights for diverse audiences. It’s a delicate balance between showcasing expertise and ensuring accessibility.

Effective communication hinges on knowing your audience. Without this understanding, even the most profound insights can miss the mark. While expertise is crucial, it’s just as vital to present that knowledge in an accessible manner. After all, it’s not about simplifying our work; it’s about clarity and connection. If we don’t make that effort, why should our audience engage?

The curse of anonymity

There is another side to this story. Have you perhaps seen something similar to the picture below?

Surely, many of us have been there. I’ve been in that spot too, with my face in the top-left corner. My record? Eight straight hours during a lengthy, and if I’m honest, tedious training I led. Other participants? Their cameras flickered on just briefly – at the start for a quick "hello" and at the end, predictably, to bid "goodbye".

From that experience, I know this:

If you treat them like a wall, they will behave like a wall.

Without engaging our audience or weaving an intriguing narrative, we can’t expect interaction in return. Most often, they’ll only engage when something annoys them. This underscores the power of data-driven Storytelling. I’m a strong proponent of this approach. Given the swift advancements in AI and LLMs, I firmly believe it should be integrated into all business-focused courses, especially finance.

The might of the overcomplication

The image above captures the entrance of an apartment building in Warsaw’s Praga district. This building was modified to accommodate individuals with disabilities, leading to the creation of this unique ramp. I’ve passed by it several times, and it always makes me wonder:

Why such a design? Couldn’t it be simpler or slightly steeper? Would that cause trouble getting in? How much time does one spend navigating from the beginning to the door? Do people use it? Would it not make sense, to buy and install a lift instead of building this ramp? So many questions, but no answers…

Initially, there were no handrails, making it look less cumbersome. But now, it’s quite the sight.

And please, please, please: don’t take me wrong. I am 200% supportive of any initiative, that makes the lives of individuals with disabilities easier. I have simply seen solutions, that in my view, could help people function the way they want, and were not as cumbersome as that thing in the picture. But who am I to judge?

Anyway, often the issue with data or analyses is, that they really can be overwhelming. They can be packed with intricate details that, while may seem important, are extraneous when it comes to actual decision-making. These unnecessary intricacies can sidetrack us, and there’s a risk that things might go awry, just like the ramp in the photo.

So what?

Let me tell you another anecdote. This time about a resourceful car mechanic from Belarus. Over his career, he amassed a heap of discarded car parts. Instead of letting them rust away, he creatively repurposed them into a towering 3.5-meter-tall dwarf. You can see it here. Quite the innovative use of "junk", don’t you think?

Isn’t that something? A ** 3.5-meter-high dwarf from discarded car parts. But for us, what’s the takeaway? Frankly, not much.** Unless, perhaps, we spot a part we recognize because it was taken from our car.

But it does not work this way in business.

Consider dashboards for a moment. Haven’t we all come across that one management dashboard that’s overloaded with data, yet offers little in terms of actionable insights? Or perhaps the valuable information is buried deep within, much like trying to spot an old carburetor inside a 3.5-meter metal dwarf. As I highlighted in a recent article, it doesn’t have to be this way. Dashboards can indeed be powerful storytelling tools [5].

Why bother?

Here’s a photo I took in my office. The note on the sheet of paper reads in Polish: "Kindly wash your dishes or place them in the dishwasher. Failing to do so will render the sink unusable for others".

Yet, you see the result.

Navigating daily corporate challenges, I’ve committed to one: washing my mug, even when the sink is brimming with dishes. True, there were situations, once or twice when the rush of the day made me skip it. True, the snapshot above showcases a state that’s about 30–40% of the fullest state I’ve ever seen in that sink.

Funnily enough, as I paused to take this photo, two colleagues wandered into the kitchen with dirty mugs. To my amusement, they too felt the urge to wash their mugs. A mere coincidence? I somehow find it hard to believe.

The power of an effective call to action is not just in stating what needs to be done, but in determining who will do it. By naming this person or team and following up afterward, we can avoid endless meetings that waste time and money. Additionally and finally, we should make use of our data!


What insights emerge from the above discussion? Most importantly, how can companies effectively transition to being data-driven?

How to dry a data swamp?

Well, the best strategy is to avoid one. The line between a streamlined data lake and a chaotic data swamp can be thin. But, with the right strategy, you can transform a data swamp back into an organized and efficient data lake.

To effectively reclaim your data lake, start with a thorough data audit to assess data relevance and structure. Introduce robust data governance, emphasizing protocols and data stewardship. Enhance metadata management for better data discoverability and implement tools that assure data quality through validation checks and automated cleansing. Periodic audits, data retention policies, and a focus on data security are essential. Embrace advanced technologies like machine learning and AI for anomaly detection and data optimization.

Equally crucial in this rejuvenation journey is building a knowledgeable team. Provide ongoing training and promote a culture that emphasizes data literacy. This holistic approach, combining strategy and technology, ensures your data lake remains an organized and invaluable asset to your organization.

How to make insights, eh… more insightful?

Now that we’ve successfully navigated out of the data swamp, it’s time to dive deeper.

For truly valuable insights from our data, we must start with the right questions [6]. Simply pivoting data aimlessly won’t yield the results we want. To pinpoint the right questions, we need individuals well-versed in the business world to guide our data use. Think of a business as a complex machine, with not only intricate inner workings but also interactions with the external world like customers, suppliers, competitors, and the economy in general.

This is where the role of the financial business partners becomes crucial. They act as a bridge, connecting financial data with real-world business operations. They guide the data teams, offering direction on what insights might be beneficial. A well-designed dashboard can be a powerful tool for them [5]. They might start by testing a few ideas on a smaller scale using such a dashboard. And when they find something noteworthy in the data, they’re the ones who ensure its significance and validity.

How much is the fish?

So, we’ve gathered our insights. But what happens if they’re unsettling? What if they expose vulnerabilities?

Leaders play a pivotal role in shaping a data-focused culture. They champion an environment of inquiry, positioning data as the solution to pressing questions. By celebrating victories rooted in data and emphasizing ongoing education, they embed data into the DNA of the organization.

Leaders lead by example, making choices that spotlight the importance of data. They establish benchmarks based on data, ensuring measurable outcomes and creating responsibility in teams. Their adaptability and dedication to data-driven methods serve as a beacon, motivating the whole organization to embrace a similar approach.

Proving the obvious…

When you’ve secured data, crafted insights with management’s backing, and narrated a fact-driven story, it can be demotivating when someone expresses skepticism, especially if others side with the doubter. So, how should you navigate this?

First, attentively listen to their reservations to grasp the nature of their doubts. Respect and acknowledge their viewpoint. Demonstrate your willingness for open conversation. Probe deeper with questions to better understand their concerns. In clarifying intricate points, lean on straightforward explanations complemented by simple analogies. It’s crucial to remain calm, ensuring the discussion stays productive.

Should skepticism persist, recommend additional meetings or workshops for a more thorough exploration. By suggesting extended engagement, you underscore your dedication to mutual clarity. Working together is better. Teamwork can help us understand more and build trust with each other.

The curse of anonymity

Oh, is there still a problem? By now, everyone should ideally be on board. If there’s a disconnect, it’s high time to harness the power of storytelling with data. Data-driven narratives, enriched by compelling visuals, offer a seamless bridge between data scientists and business teams. Rather than delving into complicated spreadsheets, concise visuals simplify the data. This ensures that the core insights are immediately evident, facilitating more informed decision-making for all [7].

Keeps things simple and to the point

Storytelling with data helps make complex things simple. By turning data into a clear story, we highlight the main points and leave out the confusing bits. Using pictures or graphs makes it even easier to understand. This way, everyone can quickly get the main idea without getting lost in too many details. In short, data storytelling keeps things straightforward and focused.

Who is taking the notes?

Finally, let’s address the burning question that inspired this article: How do we effectively utilize insights and data?

The secret lies in preparation. A compelling call to action is built on a foundation of a well-crafted story derived from data. This call to action should be unveiled right at the story’s climax, that pivotal moment when all eyes are on you (or your slides), and the audience is eager to hear what comes next. But this momentum is short. No matter how impactful your message is, it’s vital to establish responsibility. Who will take the next steps? What will be done next?

Ideally, someone in the audience will note this down (you have to ask for it beforehand). Shortly after your presentation, send out a summary email detailing the key decisions made. Periodically, check in to monitor progress and ensure the ball keeps rolling.

Concluding remarks

Every big journey starts with a single step. Even small things can make a big difference. Using data is important for today’s businesses, but it’s not always easy. There’s no one right way to do it, but starting the journey is worth it.

Follow me on medium.com.

Connect with me on LinkedIn.

Visit my blog (in Polish).

Visit my YouTube (in Polish, films with EN subtitles available).

*After deducting memory used by iOS and system files.

References:

[1] Duarte, Fabio, Amount of Data Created Daily (2023), April 3, 2023

[2] I supported myself with this article to make the above-displayed computations: Zero Agency, Counting the Zettabytes, October 21, 2015

[3] NewVantage Partners, _Data and AI Leadership Executive Survey_, January 2022

[4] Silva Santos, Diogo, The Past, Present, and Future of Data Architecture, March 9, 2023

[5] Szudejko, Michal, Leveraging Management Dashboard For Storytelling: A Viable Pathway, August 3, 2023

[6] Ramakrishnan, Rama, I have data. I need insights. Where do I start?, July 2, 2017.

[7] Nussbaumer Knaflic, Cole, Storytelling with data, Wiley, 2015.


Related Articles