How to Build a Modern Data Team Structure?

It is the 21st century and you are leading a fast-growing fintech startup that is about to hit a breaking point. The data team has doubled in size over six months, but chaos is reigning. Analysts are wasting hours reconciling conflicting reports, engineers are scrambling to fix broken pipelines, and leaders can’t agree on priorities. The result? Missed opportunities, frustrated teams, and you are asking: “Why aren’t we getting value from our data?”

A Gartner survey found that 56% of companies struggle in having effective teams to bring value to the organization, despite heavy investments. There is a large positive shift in companies today—modern data teams now drive business-critical data products, automate decision-making, and influence strategic initiatives. This evolution means data teams must balance two competing priorities: foundational work and insight generation. Failing to balance these priorities can result in inefficiencies.

We will break out how top firms like Airbnb, Revolut, and Nubank set up their data teams. You’ll get actionable frameworks, real-world examples, and a clear path to building a team that delivers results—not headaches.

Why does Data Team Structure Matter?

As businesses grow, so does the complexity of their data operations. A well-balanced data team guarantees data pipelines are strong, machine learning models are efficient, and insights are actionable. McKinsey discovered that businesses that leverage data effectively are 23 times more likely to do better than their competitors.

The Pillars of High-Performing Data Teams

Although data teams manage never-ending responsibilities, their work reduces to three fundamental functions. Get the right balance, and you will prevent chaos from impacting most companies.

1. Insights Roles: The Storytellers

“What do these numbers mean, and what should we do next?”

Organizations generate vast amounts of data daily, but raw data alone isn’t valuable—it’s the insights that drive decision-making. The insights team plays a crucial role in transforming numbers into actionable strategies.

I. Data Analysts

Turning raw data into structured reports and dashboards that provide visibility into business performance. They provide visibility into performance metrics, helping teams track key indicators like sales, customer acquisition, and marketing campaign effectiveness. Their work involves cleaning and structuring data, identifying trends, and presenting findings through visualization tools like Power BI, Tableau, or Looker. They answer questions like:

• How did our latest marketing campaign perform?
• What’s the trend in customer acquisition over the past quarter?

II. Product Analysts

Optimizing the user experience by measuring how customers interact with products and services. They work closely with product teams to assess feature engagement, identify drop-off points, and evaluate the impact of new releases. Work with product teams to maximize features (such as measuring app participation rates). They responded to inquiries like:

• Which app’s features grab users’ attention the most?
• Why are users leaving their shopping carts before they check out?

III. Data Scientists

Forecasting results and addressing difficult company challenges using innovative modeling techniques. Predictive and prescriptive analytics allow data scientists to forecast future trends and automate decision-making procedures. 

Their expertise in machine learning, artificial intelligence, and statistical modeling allows them to answer high-impact questions like:

• Which customers are most likely to churn?
• How can we personalize recommendations for users?

At Netflix, analysts discovered that 75% of viewer choices are driven by recommendations. This insight reshaped their entire content strategy.

2. Data Engineering Roles: The Architects

“Is this data accurate, and can everyone access it?”

I. Data Engineers

Design and develop pipelines that move and transform raw data. They work with ETL (Extract, Transform, Load) processes, streaming technologies, and cloud platforms to integrate and prepare data for further analysis. (e.g. syncing CRM data to a warehouse). You can use tools like Hevo which simplify data integration, enabling teams to automate pipeline creation and maintenance, ensuring data is always accessible and up-to-date.

II. Analytics Engineers

Model data for clarity. Analytics engineers optimize how data is stored and structured, making it easier for analysts and data scientists to extract meaningful insights. They create clean, well-organized tables, define business logic within transformation layers, and use tools like dbt (Data Build Tool) to standardize datasets. (e.g., using dbt to clean tables for analysts).

III. Data Platform Engineers

Maintain and scale the data infrastructure. They ensure that cloud storage, data warehouses, and processing frameworks are scalable, secure, and performant. Their expertise lies in managing distributed computing frameworks, optimizing storage costs, and ensuring system reliability.

By centralizing their engineering efforts, Airbnb was able to slash storage costs by 27% and reduce Amazon OpenSearch Service costs by an impressive 60%.

3. Machine Learning Roles: The Innovators

“Could we automate decisions or forecast results?”

I. Machine Learning Engineers

Bring AI models to life. Machine learning engineers implement models—such as fraud detection systems, deploying a recommendation engine for e-commerce, or optimizing predictive maintenance models in manufacturing. ML engineers ensure that AI solutions are reliable, efficient, and scalable.

II. AI Researchers

Explore innovative ideas, new algorithms, architectures, and methodologies to improve AI capabilities. (e.g. developing novel deep learning models, enhancing generative AI techniques, and refining natural language processing (NLP) approaches).

With ML to customize product recommendations, European e-commerce behemoth Zalando increased their revenue by an estimated 3.9 percent to 10.5 billion euros in 2024.

The Golden Ratio: How Top Teams Allocate Roles

“What is the right ratio in my company?”. Well, it depends. We first classify companies into different groups by company size:

• Mid to small growth and small size companies with a data team smaller than 35 (e.g., Typeform, Brex, and Personio)
• Mid-size scaling companies with a data team of 35–100 (e.g., Notion, Miro, and N26)
• Large enterprises with a data team of 100+ (e.g., Zendesk, LEGO, and Nubank)

We have studied some different research conducted by organizations to understand what this golden ratio is. 

• The median proportion of people in insights roles is 46%, slightly higher than the 43% in data engineering roles, with the lowest being 11% for machine learning. 
• Larger companies have a higher share of people in ML roles. They invest heavily in machine learning and governance. A global tech giant like Amazon or Google invests in machine learning teams for personalization and fraud detection while also focusing on data governance.
• Small teams prioritize engineering to build foundations. A fintech startup might first hire data engineers to ensure transaction data is structured and reliable before investing in analytics or AI models. 
• Mid-size teams need more analytical talent to generate business insights.
• Semantics play a big role as some companies prefer to call analysts as data scientists. 

An analysis of 100+ companies from US and Europe revealed striking patterns:

Proportion of Data Teams in Companies

Most companies keep data teams from 1% to 7% of the total workforce. 

Fintech companies *e.g. Revolut, Wise) come out on top with 3.5% of the workforce being in data roles while B2B companies (e.g. Figma, Talkdesk) have the lowest ratio at 2.4%. That makes sense because fintech companies are heavily reliant on data for core operations like fraud detection, credit scoring, risk assessment, and regulatory compliance. They need real-time analytics, machine learning models, and robust data infrastructure to make critical financial decisions.

On the other hand, B2B companies typically have a longer sales cycle, lower transaction volumes, and a greater focus on relationship-driven selling rather than real-time, data-driven decision-making. While they still use data for customer insights and operational efficiency, it’s not as central to their business model as it is for fintech companies.

Proportion of Data to Engineer Ratio

A deep dive into the data to engineer ratio reveals that companies in the marketplace vertical have the highest data to engineers ratio. That makes sense because marketplace companies (e.g. Uber, Airbnb, and Amazon) operate on massive amounts of real-time data, making data roles crucial. 

Case Studies: How Real Teams Get It Right

1. Revolut: Scaling Insights Without Chaos

As Revolut expanded to 35+ countries, it has invested in analytical talent to combat fraud and cybercrime. Revolut’s report divulges the growing threat of scams, which robbed global citizens of over USD $1 trillion in 2022, according to GASA (Global Anti-Scam Alliance & ScamAdviser, Global State of Scams report). Revolut’s proprietary fraud detection system, powered by cutting-edge machine learning and artificial intelligence, prevented over £475 million in fraud against customers. 

2. Airbnb: From Broken Pipelines to World-Class Infrastructure

Airbnb has a higher number of engineering roles open than analytical roles. It operates a marketplace business model, which means it heavily depends on data infrastructure, real-time systems, and automation rather than manual analysis. 

3. Shopify: Where Code Meets Commerce

Shopify emphasizes empowering merchants by building products that help them make the best decisions faster and more intelligently. Their engineering and data teams work on creating, training, and deploying machine learning models at scale. 

The 5 Unspoken Rules of Data Team Design

1. Start with Engineering (Yes, Even Before Analysts)

According to Lenny, B2B business leaders shared their first 10 hires of their team of superheroes. 70% of them hired engineers as their first number 1 employee.

2. Embed Analysts in Business Teams

Analysts should understand context and collaborate directly with business teams.

3. Treat ML Like a Product—Not a Science Project

A 2021 VentureBeat analysis suggests that 87% of AI models never make it to a production environment. Doordash invested in Machine Learning Workbench to revolutionize their MLOps. 

4. Make Data Governance Everyone’s Job

According to Gartner, “Every year, poor data quality costs organizations an average of $12.9 million.” 

5. Measure What Matters: Outcomes, Not Outputs

Remove bad metrics like “Number of dashboards created.” but instead use good metrics like “% of company decisions using data.”

The Future of Data Teams: 3 Predictions

1. Analytics Engineers Will Replace Traditional Analysts

Tools like dbt and Snowflake are democratizing data modeling. Expect “analyst” roles to merge with engineering. Nubank data analysts are now called analytics engineers. 

2. ML Engineering Splits into Two Roles

“ML Ops” (deploying models) and “Applied ML” (solving business problems) will diverge.

3. Data Teams Shrink (But Get More Strategic)

AI-powered tools will automate a fair proportion of routine tasks, letting teams focus on high-impact work.

Your Action Plan: Build a Team That Scales

1. Audit your current structure: Use the 46/43/11 ratio as a starting point.
2. Fix the foundation first: No insights team can thrive with broken pipelines.
3. Hire for hybrid skills: Look for analysts who know SQL and business KPIs.
4. Kill vanity metrics: Tie every role to a business outcome (e.g., “Reduce customer support costs by 15%”).

Conclusion

The journey to structuring a top data team isn’t about chasing trends or copying competitors—it’s about solving your company’s unique problems. From Revolut’s army fighting fraud to Airbnb’s engineering rebuild, the best teams share one trait: they align their structure with their mission.

Here’s what matters most:

• Balance beats perfection. Invest in engineering early, empower analysts to collaborate with business teams, and scale machine learning only when it drives measurable value. You can use tools like Hevo which can help streamline data integration, ensuring your pipelines are robust and reliable from the start.
• Clarity is king. Ambiguity wastes time. Define roles, set expectations, and measure outcomes—not outputs.
• Governance isn’t optional. Whether you’re a startup or a Fortune 500 company, clean data is the fuel that your company needs to function. Start small, but start now.

The future belongs to teams that stay agile. As tools like AI and automation reshape the landscape, the “perfect” structure today might evolve tomorrow. But the core principle remains: Great data teams don’t just crunch numbers—they solve real problems.

Now, go build a team that turns data into your company’s superpower—one insight, one pipeline, and one model at a time. 

Frequently Asked Questions