Building meaningful connections with loyal customers
Of retail consumer data ingested to support the AIR MILES Reward Program
Personalized offers per month powered by machine learning
In redemption value, contributing to $70 billion in total sales
Since 1992, LoyaltyOne has owned and operated the AIR MILES Reward Program, with nearly 10 million active collector accounts earning rewards through everyday purchases at more than 300 retailers, generating $70 billion in revenue across their network. With billions of data records from collectors and retail partners at their disposal, AIR MILES had an opportunity to foster stronger touch points with each of its loyal customers to improve retention and customer lifetime value. To achieve this, they needed to modernize their lagging legacy infrastructure, which struggled to provide a holistic view of their customers, preventing the company from making more intimate connections. With the Databricks Data Intelligence Platform, AIR MILES is tapping into their data across all facets of their business — including improving revenue forecasting; providing more efficient and targeted customer support; delivering personalized marketing campaigns to improve the collector experience; and serving insights to retail partners so that they can better understand how customers are participating within their loyalty programs. With unified data access and insights, AIR MILES is now maximizing its foundation to further differentiate itself and empower partners with new levels of personalization to drive significant business impacts.
Struggling to expand data capabilities on legacy infrastructure
AIR MILES isn’t tied to a specific retailer, brand group or business type. As a coalition rewards program, AIR MILES offers an open and flexible model that provides collectors with more ways to earn and redeem Reward Miles for a wide variety of redemption options through partnerships with other businesses. This untethered model has allowed AIR MILES to capture petabytes of customer data generated by 558 million transactions per year, which they wanted to leverage to enhance the collector experience and enable their partners to better engage and drive revenue. To do so, they rely on the billions of records of data they have from their collectors and partners to compile a 360-degree view of customers that can be leveraged for sales and marketing, and product and offer development.
Originally, AIR MILES didn’t have a big data presence. While they were using some of the most capable databases, hardware and analytics tools of the time, Emmanuel Menard, Director of Data Science & Data Engineering at AIR MILES, says, “We wanted to take our technology 10 years ahead. We had the vision to do everything in streaming technology without storing data — just requery the data live every time — but the tools we were using couldn’t get us there.” The company’s data scientists couldn’t access data older than 5 years, which hindered their ability to gain a complete view into the customer. Plus, complicated queries limited what they could do with the data they did have access to, and pipeline development alone took 3 to 4 months, which stifled their ability to deliver innovative solutions to market in a timely manner.
Despite their clear vision and mission to productize data, AIR MILES needed a modern approach that could unify all their data — both batch and streaming — and facilitate the building of reliable and efficient pipelines to deliver actionable insights in a meaningful way.
Lakehouse architecture empowers all areas of the business to be data driven
Once AIR MILES began to migrate workloads to the Databricks Data Intelligence Platform, the fully managed cloud-based platform quickly revealed its value through speed, efficiency and openness. Now all its data — including operational data such as website and mobile logs, email communications, sales and marketing campaigns, and transactional data from partners — is made readily available to all departments across the company. Through the delivery of business intelligence (BI) dashboards via Looker and Tableau, analysts and nontechnical users can make data-driven decisions for the business. Looker generates partner-facing reports that summarize collector profiles, reveal promotion effectiveness, and compare discount and product promotions, as well as other variables necessary for retail partners to improve their business and marketing performance. Tableau reports provide the rewards and marketing teams with insights to understand mileage accumulation and redemption trends, different types of collector profiles based on campaigns, digital and marketing insights, and ROI on campaigns.
The data science team is also benefiting from the simplified and expanded access to their data. They are now able to quickly build, test and deploy machine learning (ML) models to deliver various use cases serving a wide variety of internal teams and partners. The marketing team is using ML to better understand web traffic performance across their mobile and web platforms and how to optimize the user experience. The finance team is able to more accurately forecast revenue, and the customer care team can improve customer support through ML-powered dashboards that help them understand how agent performance metrics are tracking to their call center SLAs. And the fraud department can detect patterns and outlier behavior to ensure promotional offers are redeemed as intended.
Menard explains, “Regardless of a user’s technical knowledge, they can access, query and apply the data. From data analysts using Databricks SQL to execute advanced queries, to novice users accessing data insights within Tableau and Looker dashboards, different departments in AIR MILES and our partners can now leverage and collaborate on a unified view of our data.”
Empowering, enabling and differentiating with data
With Databricks Data Intelligence Platform as the driving force behind its evolution as a data and analytics organization, AIR MILES surpassed the initial goal of obtaining real-time data insights — fully integrating a data-first approach throughout the company. Under the hood, the key to their success is the blazing performance of the platform, which results in faster insights to the end users. Data pipeline development that used to take 3 to 4 months now takes just 2 days with Databricks. Now, AIR MILES has over 100 data pipelines that deliver faster time-to-insights with more relevant data. They have also implemented a metrics catalog framework that allows them to curate and manage all their business rules in one centralized location. This higher throughput accelerates the time-to-market of new insights and ML-powered solutions. Data teams are also empowered with 30 years of historical data at their fingertips that can be easily accessed and queried without complication.
Through this “open data market,” over 100 internal AIR MILES associates across departments and teams, plus additional users and data vendors, are now able to access data insights fed through BI tools. And with over 23,000 collector attributes at their fingertips, their data science teams have developed over 250 ML models to hone personalization for business growth. Using its product recommendation models, AIR MILES has been able to increase product sales 4x through personalized email marketing campaigns targeting 4.7 million opted-in collectors. With their reinforcement learning models, AIR MILES has also increased collector response rates to promotional offers by 2x. Through the delivery of 100 million personalized offers per month, they have driven $600 million in redemption value, contributing to $70 billion in AIR MILES sales across their network.
“You can have a million customers, but if you don’t know who they are, you can’t build relationships with them. With Databricks, we can look toward the future and identify the next big opportunity to help us build lasting and meaningful relationships with our collectors” says Paul Raso, Associate Vice President, Data Platforms & Products at AIR MILES.
