The Craft Behind The Numbers

Data science gets described as a technical discipline, but that framing misses an essential piece. The field originally emerged from three traditions: science, mathematics, and art. Over time, the artistic component has faded into the background while the technical side dominates. That's a loss, because the artistic skills—especially communication—are what turn raw analysis into something a business can actually use.

Communication is the hinge point for everything a data scientist produces. Without it, stakeholders can't understand the work, let alone act on it. Data storytelling is the mechanism that bridges that gap. It captures attention, builds trust, and gives people who might be intimidated by numbers a shared foundation to work from. When you frame your analysis as a narrative, you're not just presenting findings—you're inviting your audience into the reasoning behind them.

What It Means To Tell A Data Story

Common definitions describe data storytelling as communicating insights through narratives and visualizations. That's true but incomplete. There's an assumption that a great chart or dashboard does the storytelling work on its own. In practice, that leaves the audience to extract the meaning themselves—and many won't know where to begin.

Real data storytelling goes beyond relaying data points. It's about making sense of the world and presenting insights so stakeholders can understand and act on them. Good stories have structure, and data stories are no different:

  • The main character: The business problem is the hero. Identify it clearly, summarize what you explored, and be willing to reframe it if that leads to deeper insight.
  • The setting: Context shapes interpretation. Provide the background information that matters, but stay neutral—don't force the data into a predetermined narrative.
  • The narrator: Speak the audience's language. Skip jargon for non-technical stakeholders, and define terms when they're unavoidable.
  • The plot: Stories need direction. Recommend next steps so the audience knows what the data suggests they do next.

1. Build The Practice Into Your Team

Telling effective data stories requires infrastructure. That starts with a solid data foundation—clean, conformed data you can access confidently and move quickly on. At Shopify, that foundation proved critical during the early days of COVID-19, when the company relied on data storytelling to understand what was happening for merchants and make decisions accordingly.

That experience led Shopify to institutionalize the practice. The company formed a working group for data scientists, led by data scientists, dedicated to advancing storytelling as a craft. Members can drop in for feedback or schedule a review of a project in progress. The group offers guidance on framing, suggests angles already explored, and helps refine how findings get communicated back to stakeholders.

The goal is simple: produce the highest quality work so that when data communications reach an audience, they come with accurate data and clear guidance. That's what establishes a data scientist as a trusted partner.

2. Use Tools That Frame The Narrative

Data scientists are in the decision-support business, and that means the story doesn't end when a dashboard ships. You're responsible for transmitting the insights to your audience. That requires tools designed for the job.

Shopify has adopted Duarte's Slidedocs approach—a format that uses presentation software to create visual reports meant to be read rather than presented. Slidedocs pack dense information and visuals into a digestible format, similar to a policy brief. Useful elements to include:

  • The data question being answered
  • A description of the findings
  • A graph or visualization
  • Recommendations based on the findings
  • A link to the full report
  • Contact information for the storyteller

There's no single correct way to build a Slidedoc—it's a creative exercise shaped by your audience and what you need them to understand. The format guides stakeholders as they explore the data and form their own connection to it. It's not the only option, though. Teams in marketing, PR, and UX have strong communication practices worth borrowing from. The point is to find formats that make information action-oriented and tailored to your audience.

3. Let The Audience Participate

The most effective data stories are experiences, not monologues. Interactivity lets audiences explore different facets of the story on demand—filtering visualizations, drilling into details, and effectively co-creating the narrative. Showing beats telling when it comes to making the story land.

Shopify applied this idea with the BFCM 2021 Notebook, a data storytelling product launched for merchants after Black Friday and Cyber Monday. Merchants already had reports and analytics showing business performance, but the Notebook gave them more agency and a personal connection to their own data.

Massive datasets can overwhelm people—they may not know where to start or worry about exploring incorrectly. The Notebook provided a scaffold for that exploration. It was an interactive visual companion that let merchants dive into their sales data by products, days of the week, or buyer location. Those who wanted to go further could click on visualizations to see the underlying queries, sparking curiosity about writing their own.

The result was merchants who felt confident exploring their data and took ownership of it. When you craft a data story, ask whether the end user has opportunities to engage with it interactively.

The Craft Behind the Numbers

Data science is often framed as a purely analytical discipline, but the label undersells what practitioners actually do. Effective data work demands a balance of mathematics, science, art, and communication. Storytelling is the connective tissue that brings those elements together, letting you shape raw findings into narratives stakeholders can grasp, reflect on, and act upon.

Putting effort into how you present data pays off in two concrete ways: it builds trust with your audience and raises their overall fluency with the information you share. Creative techniques and interactive elements aren’t decorative extras—they are essential tools for making insights stick.

Why Storytelling Matters

A chart or a table can show what happened, but a story explains why it matters. When you frame data within a narrative, you give your stakeholders a way to mentally organize the numbers, connect them to business context, and remember the key takeaways long after the meeting ends.

This approach also shifts the dynamic between data teams and decision-makers. Instead of simply delivering reports, you invite stakeholders into a conversation. They can ask questions, explore underlying assumptions, and see how different scenarios might play out. That collaborative loop reinforces confidence in the data and in the team producing it.

Interactivity as a Bridge

Static visualizations limit exploration. Adding interactivity lets users test their own hypotheses and see the consequences of changing variables in real time. This turns a one-way presentation into a discovery session, where stakeholders can verify claims for themselves rather than taking them on faith.

The payoff is a deeper level of engagement. People who can poke at a dashboard and see the data respond tend to walk away with a more nuanced understanding than those who simply viewed a fixed slide. That self-driven exploration also reduces the burden on data scientists to anticipate every question upfront—users can find many answers on their own.

A Hybrid Discipline

None of this diminishes the rigor of the science underneath. The analytical methods still have to be sound, the models validated, and the measurements accurate. What storytelling adds is the layer that makes that rigorous work legible and persuasive to non-specialists.

The creative side of data science is not an afterthought to be handled at the end of a project. It deserves attention throughout the pipeline, from deciding which questions to ask to choosing how to frame the final output. Teams that treat presentation as an equal partner to analysis find their work has more impact, reaches more people, and drives better decisions. In that sense, the art isn't just nice to have—it's central to turning data into action.