AI helps create personalised, safer lottery experiences
- 4.7M Registered customers
- 11K Tickets sold per minute at peak
- 3,800 Retail outlets in network
The Lottery Corporation is the driving force behind Australia’s largest lottery games and one of the best-performing lottery businesses in the world. Every year its games create hundreds of new millionaires, delivering thrilling moments that inspire Australians to dream.
“For us in Australia, the equivalent of one in two adults purchases a lottery ticket each year. We have approximately 4.7 million registered customers, so it’s a lot of registrations and activity we get to track,” shares Stewart Campbell, head of customer analytics and data science at The Lottery Corporation.
The Lottery Corporation’s broad portfolio of games and audience demographics requires an experience-led approach using artificial intelligence and machine learning (AI/ML) to deliver personalised experiences across all channels to customers who have consented to receiving offers of this kind. While lotteries are generally considered as a very low-harm form of gambling, The Lottery Corporation has an active and overriding approach to harm minimisation, which governs its marketing approach and use of technology in this regard.
“Our number one priority of the use of AI and analytics is to give customers particular offers we think they’ll be interested in rather than a more generalised approach,” explains Campbell.
The Lottery Corporation’s AI and machine learning use cases include:
- Next best offer, churn, and engagement: To drive content and offers across channels, cross-sell opportunities, and communicate across channels and timing
- Early intervention: To help identify risk of gambling harm and drive early intervention communications
- Anomaly detection: To help reduce risk and drive investigations into activities within the retail operator network
Using AI with ClearScape Analytics™ for a next best offer recommendation engine
Next best offer recommendation engines are about giving the right product offer to the right customer at the right time. Using customer purchase behaviors, The Lottery Corporation creates personalised and cluster-based offers and communications for its c. 10 million total customers, of whom c. 4.7 million are registered.
Retail, membership, marketing, digital, and tracking systems feed data into Teradata VantageCloud on AWS, The Lottery Corporation’s Trusted AI platform. Customer, transaction, and event data powers AI/ML models using ClearScape Analytics.
“When we have enough behavioral information about a customer, we have the ability to take a personalised approach that is more aligned to the customer’s preferences and results in a more efficient use of marketing spend.”
The Lottery Corporation’s next best product and ticket offers are based on propensity models by customer, product, and day.
More than 100 features are generated across four time-series lookback windows. Teradata’s Enterprise Feature Store indexes and stores features for reusability, which improves AI/ML implementation by reducing time spent on feature engineering. This helps The Lottery Corporation scale its data science and AI efforts, as features accumulated over time can be continuously updated, documented, and used to support new machine learning and AI projects across its business.
Amazon SageMaker and Python models are trained and exported to Teradata VantageCloud on AWS using Bring Your Own Model (BYOM) in ClearScape Analytics.
“Using ClearScape Analytics for native analytical models, as well as the BYOM capability for machine learning models, means we can use the most advanced models and then deploy and score them in Teradata. We get the flexibility to use the best models but at the lowest cost possible, which is running the models in database. Otherwise, running models outside of Teradata means having a high compute cost because as models scale up, so do costs,” explains Campbell.
The Lottery Corporation is continuing to evolve its responsible gambling models, something Teradata helps make possible. Using ModelOps, The Lottery Corporation monitors and manages its machine learning models. ModelOps provides advanced model governance to help mitigate risk by automating model and data drift monitoring, simplifying model deployment, and standardising model lifecycle management.
Tailoring communications and offers across digital, physical, and out-of-home advertising has also led to increasing customer engagement and take rates.
Early intervention AI models help reduce the risk of harm and drive safer playing behavior
Creating positive impacts also means being a responsible operator and minimising the risk of gambling harm. AI and data support The Lottery Corporation’s purpose of designing and developing models that detect potentially risky gambling behaviors and intervention strategies. By designing and developing customer care interventions that help reduce the risk of gambling-related harm, The Lottery Corporation increases operator trust, customer acceptance, loyalty, and longevity and attracts new customers as it builds a sustainable platform.
Trusted and reliable data is key for The Lottery Corporation.
“We have a curated and cleansed data environment. Customer, transactional, and retail-level data points are curated into our modeling database in Teradata. We can then create a series of AI models intended to help pick out the small proportion of customers heading toward harm by applying a series of markers of harm,” Campbell says.
The output of the model helps The Lottery Corporation to make more effective and targeted interventions for the small number of players identified as showing signs of harm.
“It’s quite heartening for us to know that we’re affecting customers in a positive way,” concludes Campbell.
AI anomaly detection models help monitor compliance
A retail network of 3,800 outlets underpins The Lottery Corporation’s strong omni-channel customer engagement and is responsible for about 60% of all lottery sales.
To illustrate this, the record $200 million jackpot in 2024 saw nearly 11,000 tickets sold per minute at peak, with retail accounting for 4,300 of those tickets.
AI technology is helping The Lottery Corporation to monitor activity in its broad retail network.
Machine learning k-means cluster models are used in ClearScape Analytics and separates normal and anomalous data. A second Isolation Forest model, an unsupervised machine learning algorithm, further identifies potential anomalous data.
Potential issues identified through this approach helps guide the activity of The Lottery Corporation’s investigations team in the retail network.
Generative AI brings excitement to already thrilling gaming experience
To help maintain its leadership as one of the best-performing lottery businesses in the world, The Lottery Corporation is turning attention to the possibilities of generative AI and is keen to partner with Teradata on its ambitions.
Generative AI possibilities include knowledge-based chatbots using retrieval augmented generation (RAG), large language models (LLMs) to predict purchase behavior, and hyper-personalised agents.
“We’re only just touching on what we can do in generative AI. We want to be able to power our predictive models with transformer models, like large language models. For example, our next best offer engine could be made even better with a transformer-based model making time series-based predictions. I’m excited to see what we can do with Teradata. We want to be able to handle big levels of data. To potentially deploy LLM models within Teradata, building out RAG systems within Teradata as well—that’s the stuff we want to explore.”
Lass uns verbinden
Erfahren Sie, wie Teradata VantageCloud Ihnen helfen kann, Geschäftsergebnisse zu beschleunigen und die geschäftliche Flexibilität zu bieten, die Sie benötigen.
Unsere Vertriebsmitarbeiter sind hier, um Ihnen zu helfen.