Delivering sky-high experiences for global travelers
Virgin Australia enhances air travel from check-in to arrival with Databricks
faster deployment of ML models
increase in real-time data availability
decrease in mishandled baggage
With an ambition to be the most-loved airline in Australia, Virgin Australia is committed to creating wonderful travel experiences for their 19 million-plus customers and more than 12 million Velocity Frequent Flyer members. Historically, teams had managed information separately, but as the airline returned to profitability after relaunching as a value carrier in November 2020, they wanted to address issues they experienced with siloed architecture, inconsistent information and a constrained legacy on-premises data warehouse. By choosing the Databricks Data Intelligence Platform, Virgin Australia can now support all users, enable federated queries across legacy tools and enhance collaboration. This strategic move led to a 75% increase in real-time data availability and contributed to project outcomes that delivered a 44% reduction in mishandled bags.
Tackling data hurdles to improve personalization and more
Virgin Australia was well aware that their travelers’ expectations had evolved to prioritize personalized experiences, enhanced operational reliability and stronger sustainability efforts. These desires were driven by technological advancements, the COVID-19 pandemic and growing environmental awareness. Virgin Australia knew they could support these preferences with a strong data and AI strategy. However, the complexity of the airline’s IT infrastructure led to significant challenges with the execution of such an approach. Worse yet, the airline’s fragmented and siloed data caused commutation gaps, inconsistent information and complicated decision-making.
Virgin Australia was dealing with the limitations of legacy data infrastructure at scale. This made it difficult to craft a program that delivered a more tailored and engaging loyalty experience through its app and marketing touchpoints.
Finally, operations were constrained by the same legacy and siloed data infrastructure. The legacy on-premises data warehouse, with its limited capacity and functionality, hindered the ability to perform real-time analytics. Additionally, reliance on different extract, transform and load (ETL) tools and reporting systems across various business units led to inconsistent answers and a lack of trust in the data, further complicating the airline’s operational decision-making. Emma Taylor, Head of Technology at Virgin Australia, explained, “We needed to understand how we performed on key metrics: departures, arrivals and proportions of flights completed. We aimed to measure performance against our targets and competitors, with the end goal of having a more consistent baseline for improvements.” With these challenges, Virgin Australia recognized the need for a unified, trusted and governed data platform to support their strategic goals and evolving customer expectations.
Virgin Australia boosts productivity, real-time data with the Databricks Data Intelligence Platform
Virgin Australia has achieved a 75% increase in near real-time data availability on the Databricks Data Intelligence Platform. The airline is using AI-powered features to boost productivity, improve baggage tracking and gain actionable insights across operations. Databricks is helping Virgin Australia scale efficiently, collaborate seamlessly and enhance the customer experience.
Harnessing advanced tools for agile data and AI strategy
To tackle their multifaceted data challenges, Virgin Australia harnessed the comprehensive capabilities of the Databricks Data Intelligence Platform. This transformation was driven by the integration of several key components that collectively addressed the airline’s constraints and enabled a more agile and efficient data and AI strategy. Recognizing warehousing was a huge foundational hurdle to surpass first, Delta Lake was implemented, along with its medallion architecture that provided structured, high-quality data layers. This architecture was meticulously designed to ensure organization, accessibility and security, utilizing a tiered structure with the bronze layer serving engineering, the silver layer catering to data scientists and the gold layer tailored for data analysts and reporting. In this framework, raw data underwent a systematic process of ingestion, cleansing, enrichment and refinement. Taylor confirmed, “The Databricks Platform was open and flexible from the start, giving us one place where we can all work together. We had so many tools and functionalities to leverage and experiment with — helping everyone discover what they needed, when they needed it, from data engineers to data scientists and reporting analysts.”
Once a solid data architecture was in place, MLflow and Unity Catalog came into the picture. According to Virgin Australia, these two additions have been the most game-changing aspects of the Databricks Platform thus far. Delta Lake’s seamless integration with MLflow supported efficient data ingestion, tracking and management throughout the machine learning (ML) lifecycle. Previously reliant on SAS, Virgin Australia leveraged MLflow to revolutionize the team’s approach to ML lifecycle management. In turn, it assisted them in the creation and deployment of extensible, reusable models that facilitated solutions, like recommendation engines and propensity models, that greatly enhanced the business’ analytics capabilities.
Last but not least, Unity Catalog was soon added to provide data governance and secure access controls to MLflow and all data on the platform, optimizing data querying and analytics practices. In simple terms, it ensured that data products were well managed and compliant with information security policies and standards, improving reproducibility and transparency. The ability to federate queries across both legacy and new environments allowed Virgin Australia to bridge their data systems and teams without the need for immediate data migration. Now, Virgin Australia is leveraging their full Databricks tech stack, having plugged in tools like Power BI for dynamic data visualization and reporting. The data teams have also begun experimenting with AI-generated comments within Unity Catalog to streamline data ingestion and Databricks Assistant — an AI-powered assistant that helps with Databricks Notebooks and more — to optimize code and boost productivity. Overall, the travel industry powerhouse feels very enthusiastic about their potential after modernizing their data infrastructure, supporting downstream analytics and ML needs to deliver significant business value.
Enhancing performance and perks for customer satisfaction
The adoption of Databricks at Virgin Australia has led to the modernization of their data capabilities, putting data in the hands of the people who can get the most value from it and deliver faster insights to the business. This transformation resulted in a 75% increase in real-time data availability and a 50% decrease in data ingestion time. Additionally, the journey to an all-in-one data and AI platform — marked by a 90% faster deployment of machine learning models and substantial improvements in analytical speed to insight — has had a profound impact on loyalty and marketing operations.
By leveraging advanced analytics and machine learning, Virgin Australia can deliver tailored recommendations and suggest next best actions, promotions and services based on both segmented and customer behavior, improving engagement and fostering loyalty.
Virgin Australia’s new approach to developing data-driven insights and helpful applications faster, cheaper and more consistently is already showing a major impact. The open, flexible and powerful platform provided by Databricks has significantly enhanced data processing efficiency and decision-making capabilities across the airline for continued growth and prosperity. Looking onto the horizon, Taylor said, “Our teams are curious learners, and we are always looking for new features and functionalities from Databricks. Since we deal with vast amounts of data related to flights, bookings, crew scheduling, maintenance and weather, we are very keen to learn more about how Databricks can assist us with our GenAI model monitoring, as we continue our fruitful partnership.”
