Why the Best Decisions Start with Strong Opinions
Welcome to the first post in our new series on Opinionated Intelligence—a foundational philosophy that guides how we build systems, interpret data, and make decisions.
A Story of Opinionated Excellence: The Original Macintosh
In the early 1980s, as the personal computing era was just beginning, Steve Jobs had a radical, deeply opinionated vision: he believed that computers should be simple, beautiful, and intuitive enough for anyone to use. At the time, this idea was not only unconventional—it was considered naïve.
Most engineers were building for power users. Jobs envisioned something else. At a time when command lines dominated the landscape, he championed the graphical user interface. He insisted on beautiful fonts and refined typography. He introduced the mouse — an unfamiliar, almost alien device. And in a move that drove his hardware team mad, he insisted the Macintosh would have no fan, convinced that computers should be silent, elegant appliances.
Jobs surrounded himself with a team that shared this vision. They weren’t just shipping a product, but were making a statement. And while the original Mac had its critics, it changed the course of computing history because it was focused, bold, and unapologetically opinionated.
This is the kind of clarity and conviction we bring to how we build and make decisions at Sundial. And it’s why we believe in what we call Opinionated Intelligence.
The Case for Being Opinionated
At Sundial, we believe: the best decisions are made when you start with a strong point of view.
Opinionated Intelligence means forming clear, experience-based beliefs about what matters, and using them to guide how you build systems, interpret data, and solve problems. It’s not about stubbornness or inflexibility. It’s about focus, clarity, and reducing complexity at every layer.
In this series, we’ll walk through how this philosophy plays out across analytics, infrastructure, modeling, and AI. And we’ll bring it to life with concrete examples from our work.
Example 1: Not All Data Is Equal
When analyzing product data, do we believe all events are equally valuable? No.
We hold the opinion that 80% of analytical value comes from a focused, high-leverage subset of events and metrics. In domains like user growth, answering 80% of common questions typically involves just 50-100 key metrics.
We developed this perspective during our time at Meta. Each product area, such as Ads, Newsfeed, Games, or Videos, had a set of canonical metrics that consistently captured the majority of insights. Being opinionated about which metrics mattered allowed us to streamline upstream systems (like ETL and modeling) and build downstream tools that were simpler, more accurate, and more actionable.
Without this point of view, teams are left sifting through thousands of low-signal events, wrangling overcomplicated pipelines, and relying on tools too generic to be useful. Being opinionated acts as a filter: distilling noise into signal, sharpening focus, aligning teams, and accelerating decision-making.
Example 2: Governing Equations Drive Understanding
We believe that every business phenomenon can be described by a set of governing equations.
Whether it’s a consumer subscription product, a two-sided marketplace, or an e-commerce platform, there exists a set of core equations that govern behavior and outcomes. These equations offer a structured, opinionated lens through which to understand and investigate complex systems.
Take, for instance, the Growth Accounting Equation, which decomposes active users into retained, resurrected, and newly acquired users. Or consider how new subscriber count in a freemium product can be expressed as the product of trial starts and the trial-to-subscription conversion rate. These are not just formulas, but foundational beliefs that shape our metrics and modeling philosophy.
At Sundial, we encode these relationships into our metric stores and metric trees, embedding the governing equations directly into the analytical substrate. This approach:
Clarifies how metrics interact and evolve
Makes assumptions explicit and testable
Supports deeper investigation of phenomena
Enables the reuse of well-understood patterns across domains
Without these opinionated abstractions, our systems would devolve into a tangle of disconnected metrics, brittle models, and ad hoc reasoning. By insisting on governing equations, we impose order, enable insight, and maintain analytical rigor at scale.
Example 3: Few Schemas Power Most Analyses
Another conviction: most types of analysis rely on just a handful of core data schemas.
Cohort analysis, retention curves, transition matrices, activation funnels—despite appearing different—often depend on the same structural patterns. By identifying these reusable schemas, we can:
Automate analytical workflows
Build general-purpose tools
Scale insight generation across teams and use cases
Our opinionated stance here leads directly to more efficient and scalable analytics platforms.
Where Else Are We Opinionated?
Our philosophy of Opinionated Intelligence extends far beyond metrics and schemas. It’s deeply embedded in how we design products, build systems, and make decisions. We don’t hedge. We make deliberate, informed choices, because clarity beats optionality, and conviction beats complexity.
Here are some of the strongest beliefs that shape our work:
Guided Exploration > Full Exploration
Exploration without structure leads to confusion, not discovery. We believe that structured, reusable data models—designed for repeatable analysis—should be the foundation of every exploratory path. Great tools don’t give you infinite choices; they give you the right starting points.
There’s a Right Way to Do Every Analysis
While there may be many ways to answer a question, there is almost always a right way to answer it well. The structure of analysis, the shape of evidence, and even the visualization should serve the story with precision. We don’t guess. We guide.
Great Decisions Require Deep Understanding
You can’t make good decisions without deeply understanding the phenomenon behind the data. And you can’t understand a phenomenon without telling the right story. Data storytelling isn't decoration. It’s how understanding takes root.
Influence > Information
Providing raw data isn’t enough. To have real impact, you must influence decisions. Experts in analytics don’t just surface numbers; rather, they use them to shape direction. We build tools that support that kind of leadership.
Analytics for GenAI Companies Must Evolve with the Product
What matters in year one is not what matters in year five. The needs of a company evolves. We are opinionated about how analytics should evolve as a GenAI company matures, and we’re building systems that support that progression.
The Future of Analytics Isn’t Dashboards
Dashboards are static. The future is adaptive, intelligent systems that surface insights automatically, based on what users need, not what they click. We're designing for that future now.
AI Will Augment, Not Replace, Analysts
From schema inference to insight generation, we believe AI will transform the analytics stack. But it won’t eliminate the need for judgment. The future analyst is augmented, not obsolete.
Unified Access Is the Only Sustainable Data Strategy
Silos are death by a thousand cuts. Warehouses and lakehouses will win, because every company needs a single, connected substrate for data reasoning. We refuse to build anything that doesn’t support that.
AI Alone Can’t Deliver Insight—It Needs Expert Systems
Generative AI on its own lacks the context to deliver depth. Real insight requires expert systems that encode structure, semantics, and domain logic. That’s why we combine AI’s creativity with expert systems’ rigor: to develop real understandings.
Does Being Opinionated Work?
Yes—but only if you’re right more often than not.
To make great decisions by being opinionated, you need to be right about your strongly held beliefs most of the time. That’s why we rely on a group of experts within the company to inform, validate, and refine our opinions.
Pratiti is a world leader in user experience research
Julie leads with excellence in product design
Stanley is a pioneer in analytics thinking
Chandra brings depth in analytics vision and strategy
Anoop is excellent at architecting systems
Collectively, we are not only experts in our domains. We are seasoned product leaders. That combination enables us to form clear, useful opinions that reduce the number of incorrect and costly decisions.
Opinionated Intelligence isn’t about clinging to static beliefs. We operate with strong opinions, loosely held. We form beliefs with conviction, but remain open and flexible enough to revise them when new evidence emerges. This balance of strength and adaptability allows us to move with purpose while avoiding the inertia of outdated assumptions. We believe that what ultimately sets teams apart is the courage to have a point of view and the humility to change it.
What’s Next
In upcoming articles, we’ll dive deeper into the building blocks of opinionated intelligence.
In a world overflowing with data and options, having a strong point of view is not a liability—it’s an advantage. And when that opinion is paired with feedback, flexibility, and expertise, it becomes one of the most powerful tools for intelligent decision-making.
Stay tuned.
