Enterprise Analytics Seen Through 500 Production Streamlit Apps at Snowflake

At Snowflake, we’ve been quietly building something interesting: an internal ecosystem of more than 500 Streamlit applications serving teams across our entire organization. What started as a few experimental dashboards for one team has evolved into a comprehensive platform supporting everything from financial reporting to security monitoring to product analytics.

This article looks at what happens when you scale Streamlit across a large enterprise, the patterns that emerge, and the lessons we’ve learned along the way.

Our platform

At Snowflake, our repository for Streamlit apps tells a story of organic growth. In June 2025, over 100 contributors created or updated nearly 4,000 Streamlit apps (including dev and production).

If you want to learn more about the platform itself — beyond who uses it and how (which is the focus of this article) — check out my colleague Zachary Blackwood’s deep dive, "How We Built an Internal Library That Powers Half of Snowflake’s Streamlit Apps."

Insights

Apps cover many use cases and business functions

Streamlit’s flexibility shines when teams can adapt it to their specific domain needs. We’ve seen everything from simple lookup tools to complex multiteam analytics platforms. Here are the four teams most addicted to Streamlit at Snowflake:

Business Intelligence & Analytics
Our finance team leads the company in app count, building everything from revenue retention analysis to procurement analytics. These apps handle critical business operations like sales pipeline tracking and cost optimization.
Product & Engineering Operations
Product teams use Streamlit for feature adoption tracking, performance monitoring and incident analysis.
Marketing Intelligence
Marketing teams focus on campaign performance, lead attribution and content analytics. These apps help track everything from paid digital campaigns to website SEO performance.
Security & Compliance
Security teams have built comprehensive vulnerability management dashboards, security findings tracking and risk treatment workflows.

App creators have different backgrounds

We also observe that app creators can have very diverse backgrounds. Expectedly, data scientists and business analysts are heavily contributing to the repository. But you’ll also find data engineers, security engineers and product managers writing Streamlit apps.

Apps have all kinds of lifespans and maintenance costs

One of our most interesting observations is the bimodal distribution of maintenance patterns. Core business apps are frequently updated (e.g., product metrics or financial reporting apps). But many apps do reach a “finished” state, especially internal tooling apps, lookup tools or established reporting dashboards. And many apps are intentionally short-lived. For instance, research or analysis apps and demo apps.

Apps start small and gradually grow

Teams typically start with simple apps and gradually add complexity. The progression from single pages to multipage applications with shared modules happens naturally as requirements grow.

Shared infrastructure, code and documentation matters

Our repository has global shared modules that handle common patterns, like CI/CD workflows, database connections, formatting and UI components. This shared foundation dramatically speeds up development, while adding more responsibility to the maintainers: We document things properly, run office hours and avoid breaking changes. That being said, all teams have their own utilities and preferences for how apps work and what they look like.

Conclusion

Enterprise analytics thrives on diversity, not standardization. Let teams experiment freely with shared infrastructure. Many of Snowflake’s most popular apps started as quick prototypes that naturally evolved into production tools. Success came from providing common utilities (database connections, formatting, CI/CD) while letting 70-plus teams build wildly different solutions for their specific needs.

Our biggest learnings:

Apps have different lifecycles — some need constant maintenance, others run stable for months/
Gradual sophistication beats big upfront design.
App creators are diverse — not at all exclusively data scientists.
Apps can be useful for every business area.

Finally, a strong enterprise analytics platform isn’t one perfect system — it’s infrastructure that enables hundreds of imperfect-but-useful systems to flourish.