How Brilliant Earth Drives Data Democratization with Snowflake and Posit

Every day, Brilliant Earth, a luxury fine jewelry and engagement ring company with a strong ecommerce footprint and a growing U.S. showroom network, automatically reprices its vast inventory of diamonds. This pricing model runs directly inside Snowflake — an automated capability that led Brilliant Earth to adopt Snowflake in the first place.

This successful automation raised the bar for the rest of the business. As Brilliant Earth expanded across channels, the appetite for insight grew faster than traditional reporting could deliver. Carefully built weekly data science reporting captured moments in time, but the team saw an opportunity in speed to delivery: What if decision-makers across the company had direct, self-serve access to the same trusted data?

“Brilliant Earth already had a highly capable data science team and well-governed data,” observed Natallia Drobau, Senior Solutions Advisor at Posit. “The opportunity was to extend that team’s reach, so their models and insights could power decisions across the entire business instead of flowing through a single reporting channel.”

Matthias Mueller, Senior Director of Analytics and Data Science at Brilliant Earth, saw that internal talent as the exact reason to change the company’s approach. "We have incredible talent at Brilliant Earth. Some of the best people I've ever worked with work here," Mueller said. "It was really an exercise in trying to bring people into the same governed ecosystem and the same platform to harness all of that talent."

To close the gap between data science and real-time execution, Brilliant Earth turned to Snowflake and Posit. Posit is the software company behind widely used open source data science tools such as RStudio, Shiny and Quarto, as well as Posit Team, an enterprise platform for building, publishing and sharing data science work in Python and R. Let’s explore how Brilliant Earth deployed Posit natively inside Snowflake to transform its central analytics function from a transactional reporting desk into a strategic builder of self-service data products.

Eliminating governance bottlenecks at the infrastructure level

In many retail enterprises, data democratization stalls because teams adopt their own isolated tools, resulting in conflicting reports and administrative nightmares for IT.

Brilliant Earth solved this governance dilemma by installing Posit Team as a Snowflake Native App running on Snowpark Container Services. Instead of maintaining a separate server with redundant security protocols requiring maintenance and oversight, the company integrated Posit directly into its Snowflake environment.

User provisioning is now entirely automated. An employee requests Posit access by clicking a tile in the IT ticketing system. The requester's manager and the analytics team approve it, and the system automatically assigns the user to the correct single sign-on (SSO) provider's group, which triggers provisioning of the correct Snowflake roles and Posit Connect account settings.

"The beautiful thing you get with deploying Posit inside the ecosystem of Snowflake is that the controls we implement in Snowflake still govern all of the access, all of the user roles, all of the tables someone can access," Mueller explained. "Posit Connect natively infers permission structures from Snowflake, so it allows our IT team to be in charge of permissions at a top-level, governed via our SSO provider," he noted, adding that this governance works seamlessly out of the box.

"Having deployed Posit on separate stand-alone servers at previous companies, moving to a native app inside Snowflake was a shift in architecture," Mueller noted. "But the seamless governance makes it an incredibly powerful system. I have not regretted deploying it this way for one minute, and I would absolutely do it like this again."

Moving from static PDFs to interactive scenario planning

Even before the transformation, Mueller’s data science team at Brilliant Earth used sophisticated data science models to optimize its investment choices. One such example is the company’s proprietary Bayesian marketing mix model (MMM). Before marrying Posit with Snowflake, these models were accessible mostly to the data science team itself; business leaders were provided with summary statistics, PowerPoint decks and other static recommendations.

To put the power of the MMM directly into the hands of decision-makers, Mueller’s data science team grounded the model in data from Snowflake and built an interactive Streamlit application hosted on Posit Connect on top of it. Today, Brilliant Earth's growth team can independently explore response curves, evaluate channel performance, set budgets, apply constraints such as channel saturation points and run custom scenario simulations complete with projected revenue and ROI metrics.

"The best domain expertise needed to actually run these scenarios lives with our business teams," Mueller noted. "By building this frontend interface, we aim to put the power directly into the hands of the decision-makers, allowing them to interact with the data model without being required to write code." The impact on the growth team has been profound, according to Mueller.

Modernizing operations for 100+ showroom managers

The next major test involved transitioning over 100 showroom operations users — responsible for tracking store-level performance — off a legacy BI platform. While the move generated significant cost savings by shedding old software licenses, driving adoption was also an opportunity to create a superior user experience.

"We're all creatures of habit," said Mueller. "Ultimately, it's about building something that someone perceives as an improvement over what they have."

Rather than simply re-creating familiar static tables, the analytics team wrapped the new Posit dashboards in querychat, a governed conversational interface. Instead of navigating through complex filters, a showroom manager can now submit a natural language prompt such as "I'm looking for the last 14 days of bookings performance for my showroom" and instantly retrieve a customized view.

Generating trusted narrative reports with AI

To replace manual weekly reporting without sacrificing mathematical accuracy, Brilliant Earth established a strict separation between quantitative calculation and text generation.

Python and R code compute the necessary metrics directly against Snowflake data. Only after the numbers are validated are they passed to an LLM’s API to summarize trends and write weekly reports in natural language.

"We use the LLM to summarize in prose, and it's really good at that. But I want mathematical functions to own the computation. What we put on Posit Connect is code-based, which means we can verify it; it’s repeatable, maintainable and reliable," Mueller said.

Accelerating rapid prototyping and disposable analytics

With AI-assisted coding and Posit running seamlessly inside Snowflake, custom application development has become incredibly fast. The team can now afford to build lightweight analytical tools designed for immediate, short-term needs.

"Having a Shiny dashboard, hosting it on Posit and connecting it to Snowflake is now so quick that I have really bought into the idea of generating disposable artifacts," Mueller explained. "In most cases, they explain the problem or analysis much better, and it is possible now to spin them out without having to spend hours and hours building them."

At Brilliant Earth, domain experts are actively prototyping their own solutions as well. By connecting frontier models such as Claude to Snowflake CoWork agents built on governed semantic layers, a merchandiser can prototype a custom report tailored to their specific metrics, such as carat weights and sell-through rates. If a prototype proves broadly valuable, the analytics team rebuilds it on validated Snowflake data and publishes it to Posit Connect as an enterprise-grade application.

Scaling governance across the enterprise

As Brilliant Earth looks to onboard its merchandising, digital and finance teams next, the goal is reinvention rather than replication as the company imagines its move into the next phase of business analytics. 

"I don't want to rebuild an ever-increasing number of dashboards," Mueller stated. "I just want to figure out what someone would like to have, work with that stakeholder and then build those things as completely custom solutions instead. We're trying to build something we want to use two years from now."

The approach reflects a crucial lesson for modern retail data strategy. As Sem Sergunin, Retail and Consumer Goods Industry Lead at Snowflake, warned, deploying AI agents on raw, unprepared data only serves to "scale your mess." Brilliant Earth has avoided this trap by establishing governed data in Snowflake, giving data scientists a robust environment in Posit right next to that data and putting interactive tools directly into the hands of business experts.

See the full story

Watch the on-demand webinar "Inside Brilliant Earth's AI Transformation With Snowflake + Posit" to hear more from Matthias Mueller and Natallia Drobau about Brilliant Earth’s data transformation with Snowflake and Posit, and watch them demo a marketing mix scenario planner. To bring custom, governed analytics apps to your own business teams, explore the Posit Team Native App on Snowflake Marketplace, where you can get started with a 30-day free trial.