How One Google Team Built Storytelling Into Analytics

News

1. Agentic AI Set to Reshape 40% of Enterprise Applications by 2026, new research finds

2. Abu Dhabi Launches Quantum-Secure Communications Testbed to Future-Proof Digital Infrastructure

3. Google Introduces AI Mode in MENA for Complex and Multimodal Search

4. How Intelligent Choice Architectures Are Redefining Decision-Making

5. AI Megaprojects Push UAE and Saudi to Global Frontlines in AI Development

6. DXC and Boomi Join Forces to Supercharge Agentic AI Adoption at Scale

EVEN THE MOST ADVANCED analytics models can fall flat if they don’t use the language of the organization’s decision makers. And when the related decision-making process stalls, it’s often because the data insights lack a clear narrative, business context, or connection to what executives care about. Those are lessons we learned at Google’s Small and Medium Business (SMB) division when my analytics team built a sophisticated model to optimize staffing for the company’s global support organization.

The model was able to forecast volatile demand across more than 100 countries by simulating thousands of possible scenarios and could recommend sales and customer support staffing levels. It accounted for seasonality, geographic differences, and even complex customer prioritization rules. We validated the data, vetted assumptions, and pressure-tested the logic.

But when we presented the model to senior stakeholders, they showed little enthusiasm. Instead of appreciating the model’s complexity, the stakeholders focused the discussion on the practicality of our recommendation model. One leader asked, “What does this mean for next quarter’s staffing in Latin America?” Another questioned how the recommendations would move the needle for her bottom line. Weeks of work stalled. No business decision was made.

This wasn’t an isolated incident. Repeatedly, technically sound models had failed to generate movement at the executive level. While many of our data professionals believed that better models would lead to improved business outcomes, executives were overwhelmed by the complexity and skeptical of black-box insights they could not contextualize.

The lesson was clear: Analytics must be built for how decisions are made, not just how data is analyzed. This required that we rethink the analytics stack — not just the data and models but also how storytelling can guide every stage of the process.

So in 2023, we began building an original framework with narrative at its core, with the goal of enabling business leaders to make faster and more confident decisions. The framework drew on lessons from a series of internal projects. Once I had conceptualized and structured the model, it was refined and pressure-tested through mid-2024 with my analytics team and business stakeholders. Since then, as head of analytics for Google’s SMB division, I’ve implemented this approach directly in high-impact projects involving sales strategy optimization, business planning, and executive decision enablement.

---

A Framework for Narrative-Driven Analytics

In traditional data analytics frameworks, the analytics team often does storytelling work after the data has been processed, or works it into dashboard formatting just before delivering insights to business users. But that’s too late. This four-layer analytics framework integrates storytelling directly into the design of the analytics stack. The framework starts with the foundational data layer, and then each layer builds on the one below it.

---

In this framework, each layer addresses a specific source of potential failure. Together, these four layers develop analytics that are based on how likely executives are to consume and act on data insights.

• Data layer: Builds business-aligned data foundations.
• Analytics layer: Designs flexible, interpretable models that anticipate and address executive needs.
• Decision layer: Structures outputs around actionable levers, trade-offs, and constraints.
• Narrative layer: Delivers coherent, contextualized insights that lead to action.

Each layer builds on the one below it to form a structured approach that helps executives make better, faster decisions.

Data Layer: Start With Shared Definitions

One of the biggest blockers to actionable insights is inconsistent data definitions. When teams don’t agree on the definitions of key business metrics, analysis becomes noisy and reduces decision-making confidence.

One project involved analyzing sales engagement to determine the ideal number of customer meetings that a sales rep could sustainably manage. This insight played a key role in informing sales coverage models and capacity planning. We discovered that teams had varying definitions of what did and didn’t qualify as a customer meeting. Sales teams defined it as everything from a brief email exchange to an all-day conference, while the business strategy team counted only scheduled phone calls or in-person conversations. We found instances where teams reported having had more than twice the number of meetings as other teams within the same region because they used different definitions. This discrepancy resulted in wildly different customer-to-representative ratio recommendations.

We quickly realized that without a common, business-relevant definition for “customer meeting,” our insights were not credible enough to support actual decision-making.

To address this, we aligned our data governance logic with business context. Partnering with cross-functional teams, we defined “customer meeting” based on the specific decision at hand. We agreed to prioritize building substantial, enduring customer partnerships. Only scheduled phone or in-person conversations qualified; emails and similar interactions were recategorized as light-touch engagement. We also built a unified data source that categorized meetings by channel and tagged each one by engagement depth.

These efforts gave business leaders confidence in the data and clarity on what it represented. Reviews required fewer clarifying questions, which allowed teams to move faster during go-to-market planning. As this example demonstrates, consistent data is a critical storytelling enabler and the first step toward action.

Analytics Layer: Build Models That Executives Understand

Analytical sophistication and accuracy are valuable, but business executives prioritize model explainability and practical relevance. Black-box models rarely earn colleagues’ trust. To secure leadership buy-in, teams should focus on transparent, flexible analytics techniques that align with how business decisions are actually made.

With a consistent understanding of customer meetings in place, the next step was to determine the optimal customer-to-representative ratio to inform our global sales capacity planning. We knew from previous experience that a static model would not be enough. Executives wanted not only the numbers but also the ability to understand how assumptions were influencing the results and how they shaped different business scenarios.

We began by identifying the scenarios that were most relevant to executives. In a cost efficiency scenario, we focused on maximizing the ROI per representative. In a growth-first scenario, the goal was to maximize overall revenue coverage. These scenarios defined the model’s structure and boundaries.

We made key drivers — like meeting frequency by customer segment, meeting length, and available rep time, after accounting for internal responsibilities like training — transparent from the start. Instead of keeping these assumptions in the background, we developed an interactive model that allowed executives to adjust inputs in real time.

For instance, executives could evaluate the impact of asking reps to increase customer-facing time by 5% or allocate more time to high-value customers. The model would recalculate the impact in real time, showcasing the downstream effects on staffing, coverage ratios, and projected revenue.

We also prioritized explainability. When results shifted, the model highlighted the most significant drivers, such as changes in customer engagement time or regional meeting norms. This gave executives a clear view into the logic and allowed them to evaluate decisions with confidence and context.

Applying this dynamic model changed the nature of our planning conversations. Debates over model assumptions gave way to discussions about trade-offs, resourcing priorities, and go-to-market execution. It reinforced a key principle: Business analytics must be decision-driven, interpretable, and context-specific to be truly effective.

Decision Layer: Ground Outputs in Real-World Constraints

Business analytics projects often don’t lead to action when findings are disconnected from the realities of decision-making. Even the most accurate and interpretable models generate results that are of limited value if they don’t reflect what realistically can be changed.

This became especially clear during our global sales capacity planning project. After building an interactive model to calculate optimal customer-to-rep ratios, we discovered that some recommended staffing moves were not feasible due to legal and operational constraints. In certain countries, labor laws required representatives to be native speakers, while other countries imposed limitations on role reassignment. These constraints were complex, country-specific, and often subject to change. Including them directly would have significantly reduced the model’s flexibility and clarity.

Instead, we revised how the model presented its outputs. Rather than suggesting that the model alone could inform staffing decisions, we identified which levers were operationally feasible and which were constrained by regulatory limits. We included callouts that clearly marked where labor policy or regulatory limitations applied.

This approach streamlined the model and grounded planning in reality. Discussions shifted from questioning recommendations’ feasibility to optimizing them within known boundaries. Because the constraints were already defined, business leaders no longer asked, “Why can’t we shift more reps into that market?”

The way decisions are framed is just as important as the model itself. Models do not have to capture every edge case, but they must clearly define what’s possible, what’s not, and why. That is what transforms analysis into action.

Narrative Layer: Lead With a Business Story

Even with clean data, sound models, and well-defined decision levers, poor storytelling can dilute impact. The final layer of the framework focuses on delivery and how to communicate data findings to enable decisions.

Rather than treating storytelling as a post-analysis presentation or dashboard polish, we embedded it into the structure and delivery of the analysis itself. In the sales capacity planning project, we built the first three analytics layers by starting with a clear problem statement, highlighting decision principles, identifying constraints, and introducing realistic, data-supported scenarios.

Then, instead of walking through technical logic from the bottom up, we led with the business context and the decisions at stake, such as how a 10% increase in sales team efficiency could affect coverage and staffing. We also used visualizations to surface trade-offs and structured our content to build executive confidence, from high-level implications down to supporting inputs.

Storytelling was not an afterthought; it was a decision enabler. By framing the analysis through a clear, compelling narrative, we helped executives not just understand the outputs but also act on them with clarity and confidence.

At Google SMB, this framework has been applied extensively, particularly in projects involving global sales coverage, resource planning, and go-to-market strategy. The framework’s principles have also been adopted by multiple cross-functional teams, including parts of Google’s finance and marketing organizations.

In high-stakes decision-making, accuracy and technical sophistication alone aren’t enough. To realize impact at the executive level, analytics must reflect how decisions are actually made, factoring in ambiguity, practical limits, and trade-offs.

Our analytics framework achieves that through:

• Clean, consistent data that creates a shared understanding.
• Flexible, transparent models that facilitate strategic exploration.
• Decision framing that clarifies what is and isn’t actionable.
• Narrative delivery that mobilizes leaders from insight to action.

By treating storytelling as a core design feature, we shifted analytics from output generation to strategic enablement. The result was not only better models but also faster alignment, clearer decisions, and more confident action.

The takeaway: When an organization’s analytics are built with decision-making in mind, analytics earns a seat at the table where it truly matters.