Thailand
PTT Oil and Retail Business (PTTOR) is a state-owned, Thailand-based petroleum and retail company that generated more than $16 billion USD in revenue in 2024. Serving nine million customers across 8 countries, including Thailand, The Philippines, Singapore, and China, PTTOR operates a diverse portfolio spanning oil and gas as well as retail brands such as Café Amazon and Jiffy.
Improving customer satisfaction and unlocking revenue opportunities
To better serve its growing customer base, PTTOR set out to deliver highly personalized marketing campaigns tailored to each customer’s individual retail needs. While its existing on-premises platform provided strong data governance, it could only support personalization for up to one million customers, limiting the company’s ability to scale.
A central part of this initiative was the introduction of “Next Best Offer” (NBO) capabilities, which use customer purchase history to recommend relevant products, such as beverage, fuel, and automotive services. However, delivering these offers required near-real-time access to customer data, which PTTOR’s legacy infrastructure was not designed to support.
As data volumes continued to grow, processing delays increased, making it difficult to scale analytics and integrate AI and machine learning into the environment. To overcome these challenges and support future growth, PTTOR turned to Cloudera to modernize its data platform and enable a flexible path to the cloud.
From 33 hours to 4: 8× faster data processing at enterprise scale
With the assistance of Cloudera Professional Services, PTTOR modernized its digital infrastructure by adopting a hybrid cloud model that utilizes Cloudera on AWS. This Open Data Lakehouse architecture enables PTTOR to process sensitive data on-premises while "bursting" critical, compute-intensive workloads to the cloud.
The solution integrates several key AWS and Cloudera services:
Infrastructure & Orchestration: PTTOR begins by processing customer data on‑premise using the Cloudera platform. This includes both transactional information—such as average spend, purchase frequency, and customer lifetime value—and demographic attributes like age, gender, and location. Predictive outputs are also generated on‑prem, including churn likelihood, estimated wallet size, and brand affinity.
Data Ingestion & Flow: All processed and modeled data is replicated from the on‑premise Hadoop environment to the cloud using HDFS and Replication Manager. Once in AWS, the data lands in Amazon S3, where it continues through cloud‑based analytics pipelines. Cloudera AI is then used in the cloud to refine, build, and deploy Python-based machine learning models that further analyze and score customer behavior.
AI, Personalization & Governance: Using these ML models, each customer is assigned their top three personalized campaign offers, based on profile attributes and predictive behavior insights. These personalized outputs are then transmitted back to the on‑premise environment and integrated into the marketing automation system for targeted campaign delivery. Throughout the entire workflow—on‑premise to cloud and back—Cloudera’s platform provides consistent security, governance, and seamless data movement.
PTTOR begins with data processed on Cloudera, including transactional and demographic information from the in-store coffee shop and membership cards, to generate predictive outputs for its customers, thereby enabling rapid development of next-best offer (NBO) proposals.
PTTOR initiated the Next Best Offer (NBO) model on Cloudera AI, where more analytics models are now deployed on premises. Additionally, PTTOR is using Geo Analytics by leveraging third-party data from the Cloud and building Geo Analytics models with Cloudera AI on Premise, which are shared via API. The data flows from on-premise systems to AWS, which is utilized for infrastructure, including Amazon S3 for data storage, Amazon EKS, and other platform services, enabling high-speed processing. The resulting predictive outputs are sent back to PTTOR’s data center for integration into marketing automation systems like Salesforce.
Quantifiable Impact: 9x ROI and Enterprise-Wide Personalization
The transition to a hybrid architecture delivered immediate and transformative financial and operational results for PTTOR’s retail business:
Exponential Scalability: PTTOR successfully expanded its NBO campaigns from a limited subset to its entire base of 9 million customers worldwide.
Massive Efficiency Gains: Data processing times were slashed by 8x, dropping from 33 hours on-premises to just 4 hours in the cloud. This reduction met the critical business requirement for daily insights and enabled greater agility in deploying new features.
Direct Financial Growth: The implementation of AI-driven personalization led to a 1.9% increase in campaign conversion rates. In 2024 alone, the NBO campaigns generated $0.81M USD in revenue growth, resulting in a nine-fold (9.1x) return on investment for the Cloudera platform.
“PTTOR receives consistent data management, strong security, and the agility to innovate faster thanks to Cloudera,” Mr. Pakorn Suriyabhivadh,Senior Executive Vice President from PTTOR said. “Cloudera simplifies our workflows, accelerates innovation, and enables us to scale AI across the business without losing control.”
Beyond Retail: Optimizing the Future of Energy
The power of the Cloudera and AWS hybrid platform extends beyond marketing. PTTOR utilizes GEO‑spatial analytics to identify high‑potential locations for new fuel stations, enabling faster, data‑driven investment decisions while reducing reliance on time‑consuming site surveys. This approach improves capital efficiency, minimizes environmental impact, and ensures sustainable development. The same capability is extended to guide the expansion of EV charging infrastructure, supporting PTTOR’s long‑term ESG and low‑carbon mobility goals.This comprehensive digital transformation earned PTTOR recognition as an EVOLVE25 Data Impact Award Finalist and Winner.
PTTOR’s hybrid data strategy acts like a high-speed metropolitan transit system: while the on-premise data centers serve as the secure main stations for core operations, Cloudera cloud acts as the express line, allowing the company to move massive "passenger loads" of data at eight times the speed of its old local tracks, ensuring every customer reaches their destination with a perfectly timed, personalized offer.
Want to see where it all began? Explore the original customer story here.
