Processed per day
THN is an innovative data and technology platform operating within the hospitality industry. Using data they collect from hotel websites around the world, THN provides hoteliers with an ecosystem of tools to help them increase their direct bookings. The tools include Oraculo, a Predictive Personalization AI model to help hotels optimize conversions on their website (“turn lookers into bookers”) and BenchDirect, an interactive benchmarking platform so hotels can compare their direct channel performance to their competitors in real time.
THN anonymously captures every event of the online user journey for more than 10,000 hotel websites across 100 countries, processing tens to hundreds of millions of data points every single day. In this case study, we explore how they use Tinybird to power the realtime metrics for BenchDirect.
THN has an average of 4 to 5 Million unique visitors per day across their hotel portfolio and multiple events and data points stored per user. One of the team’s biggest challenges is not just how to capture and store all this data, but also how to exploit it in real-time via BenchDirect so that their users have access to relevant, up-to-date competitive insights.
To power BenchDirect, THN anonymously retrieves and stores almost every single event and relevant data point for each visitor to a hotel website. The data points include each price shown to each visitor for each property, the properties they visit, offers they see, searches they make, bookings they make, etc.
Initially, THN stored all the data in MySQL and tweaked it to the limit with pretty impressive results. However, with ever-growing volumes of data and customers, they found themselves constrained by complex, long-running and difficult-to-monitor data pipelines, slow queries and slow-to-render charts and analytics.
Today THN continuously ingests data into Tinybird (initially via CSV files, now via Kafka). This data is automatically stored and materialized as needed in such a way that it is ready for consumption from BenchDirect. They have more than 30 Tinybird API endpoints serving all requests and powering BenchDirect metrics, as well as over 140 pipes that materialize data on ingestion.
Through Tinybird’s advanced templating mechanism that facilitates creating dynamic endpoints, BenchDirect can display all the necessary tables and charts which require dynamic statistical computations, filtering and aggregation capabilities.
When it comes to creating complex queries, thanks to Tinybird pipes, THN’s development process is now faster and more maintainable, since they can easily break down problems into smaller steps. And those same Pipes can then be easily published as REST APIs.
In summary, THN has integrated Tinybird as part of their normal development process, and is now able to power BenchDirect without having to worry about scale, enabling their customers worldwide to dynamically compare their hotel performance to their competitors in real time.