At Snowflake Summit '26, Chris de Groot, Manager of Data Engineering Customer Service, and Jay Stricks, Group Product Manager, Insights Platform, took the stage to share Booking.com's massive data transformation.
In their session, "Booking.com's Data Travels: Platform Foundations to Agentic Analytics," they laid out a masterclass on how to make a colossal, fragmented data landscape entirely AI-ready.
At the heart of their presentation was a core realization: AI doesn't replace the need for data architecture—it actually raises the bar.
Here’s how Booking.com moved past standard dashboards to establish a true, single source of truth using ThoughtSpot and Snowflake.
Booking.com operates on a massive scale. Their data landscape involves hundreds of thousands of sources, tens of thousands of workflows, and thousands of tables powering enterprise decision-making.
Historically, accessing this data relied on two primary vectors:
Limited, static guided analytics dashboards
Human data analysts pulling specialized reports on demand
With more than 2,000 dashboards in active use, the system simply wasn't scalable. Relying heavily on individual specialists to build custom on-demand reports created a massive data bottleneck, leaving business teams waiting for critical, time-sensitive insights.
The team was ready to move beyond dashboards to a new model of trusted agentic intelligence that delivered conversational experiences. The first step in this transition was shifting how their data was structured and governed.
Chris de Groot compared data warehousing without a model to building a house without a design blueprint, for which Booking.com has established Snowflake at the core of its cloud data platform.
To enable business users to ask complex questions, like "Why did APAC conversion drop 8% last Tuesday for family-friendly hotel properties?" Booking.com built a strict four-step pipeline:
Step 1: Clearly define business concepts and key metrics.
Step 2: Build governed, standardized tables using tools like dbt over raw data layers.
Step 3: Provide a consumer-ready interface that integrates metadata, business entities, and metrics.
Step 4: Deploy final data products to end-user applications.
De Groot shared how by connecting the Customer Service (CS) Data Warehouse Data (DWH) Model 1:1 into ThoughtSpot, Booking.com effectively removed the data analyst bottleneck.
"When we introduced ThoughtSpot, we immediately saw our backlog dropping and our analysts using the liveboards on ThoughtSpot to find ways to drive our business forward."
Chris De Groot
Manager of Data Engineering Customer Service, Booking.com
Because ThoughtSpot acts as a semantic layer, all definitions and relations are consistently reused. This secures the CS DWH as the absolute single source of truth. Business users and data analysts now have a marketplace of AI skills tailored to their roles:
Business Users: They can simply use natural language tools like ask-data or anomaly-why to automatically generate reports, look into metric drops, and explain discrepancies without writing a single line of SQL.
Data Analytics Engineers: A critical new role introduced to maintain this framework—focus their energy on managing the foundational data models rather than fielding endless ticket requests for manual data pulls.
1. Liveboards: The next evolution of interactive dashboards.
2. KPI Watchlists: An easy way for business operators to instantly keep track of performance metrics.
3. ThoughtSpot Mobile App: Enabling insights from any place, anywhere.
4. Spotter, deep reasoning AI analyst: Serving as an autonomous assistant for deeper analytical dives.
5. Answers: For deeper granular drill-downs when specialists need to dive into the raw mechanics of an anomaly.
Booking.com's story proves that the right AI-ready architecture is what makes true analytics self-service possible. With governed, standardized data at their fingertips, anyone at your org can find meaningful insights on demand.
Want to see firsthand how ThoughtSpot can help your team remove analytics bottlenecks? Schedule your personalized demo today.
The most effective way to eliminate an analytics bottleneck is to remove dependency on individual data analysts for routine reporting. This means building a governed, standardized data model and deploying a semantic layer that lets business users query data directly in natural language—without writing SQL or waiting on a specialist.
An AI-ready data architecture starts with clearly defined business metrics and concepts, governed tables built on standardized raw data, and a consumer-ready interface that exposes those metrics to end users. Booking.com achieved this by combining Snowflake for data warehousing with ThoughtSpot as the AI-powered analytics layer on top.
Empowering non-technical users requires two things: a clean, trusted data foundation and an interface that speaks their language. Natural language query tools, anomaly detection, and role-based AI skills mean business users can answer their own questions — like why a regional conversion rate dropped — without any technical expertise.