Case study: Design and deploy an open source data analytics platform

Travelport Digital is the world’s largest dedicated mobile travel technology provider, delivering solutions that allow airlines and travel companies to interact with their customers through sophisticated, realtime and personalised services.

The opportunity

Travelport was seeking to build an analytics product to complement it’s core product offerings and give its customers a deep understanding of end-traveller behaviour throughout the travel lifecycle. The product would include an extensible data platform, feeding a range of analytics applications and techniques.

• The data platform would capture travelers’ interactions with mobile travel applications related to everything from research & bookings to itinerary management and ancillary purchase patterns.
• The analytics capabilities would stretch from traditional business intelligence reporting, to investigation, prediction, and the ultimate optimisation of mobile travel applications, using machine learning and AI techniques to deliver the best possible outcome of each user interaction.
• A flexible architecture was needed, to support evolving business requirements, and capable of scaling to meet a growing customer base with huge data volumes.

The cooperation

Travelport and Sonalake employed a highly collaborative and agile project methodology to design, develop and deploy the solution based on leading open source software technologies, embedding Looker as the first analytics application for reporting and business intelligence.

The following phases were defined to manage the project complexity:

• Requirements refinement and roadmap prioritisation
• Data platform architecture design and rapid prototyping
• Evaluation & selection of reporting & BI application
• Design of event model and integration SDK for Travelport applications
• Release 1.0 system build, integration, deployment and benchmarking

This consultative and iterative process allowed for refinement of end-user and technical requirements, including an advanced degree of cloud automation to enable production support.

“As a co-development project, the cooperation between the parties was fully transparent and the project was delivered within budget and on time, with fortnightly sprints, demonstrations and regular retrospectives ensuring the final solution hit target for end-users, business and technical stakeholders”

The solution

• Architecture designed to allow individual elements to evolve over time and to scale to meet growth projections
• Built using current best in class open source software components
• Supports real time streaming and batch processing
• Provides data lake for long-term event storage
• The first reporting application uses Looker to provide flexible customer-facing dashboards, visualisations, and ad hoc data exploration capabilities
• Amazon Web Service (AWS) multi-tenant deployment with auto-scaling

The Looker selection

The initial requirement for embedded analytics functionality within the Travelport solution centred on reporting and data exploration, enabling analysis of end-traveler usage for the full range of mobile travel applications provided to airlines and travel management companies.

Based on the requirements, Sonalake undertook an analysis of leading commercial and open source business intelligence solutions in collaboration with Travelport. The analysis included hands-on evaluations of four shortlisted products, resulting in Looker being selected as the optimal choice.

Looker scored highly on technical, operational and commercial considerations.

The support for a truly embedded solution within Travelport’s product range was a distinguishing factor, with a user friendly modern look and feel aligned with the high standards of UX in Travelport’s applications.

The data modelling language LookML is an efficient enabler for definition and integration of multiple data sources, removing technical barriers for business users to explore data through report definition, data drill-down and ad-hoc queries. Dashboarding features, visualisations, security, tenancy support and scalability were all important considerations for the final selection, and the subsequent support from Looker during the implementation phase was collaborative and contributed significantly to project success.

The result

• The business benefits delivered to Travelport’s airline and Travel Management Company customers is the ultimate measure of this major development project
• Easy self-service access to advanced intelligence on the lifecycle engagement of the end-user traveler base has been created, capturing interactions from reservations to in-app promotions, rebooking in times of delay and inflight entertainment
• Together with real time analysis of clickstream data, all raw and aggregated data is now available for trending and deep learning analysis by current and future machine learning applications, addressing opportunities such as segmentation and recommendation to create the best outcome for every user interaction with a Travelport application