Sales Automation
TPBank had modernized its data estate on IBM Cloud Pak for Data, but a digital credit-card acquisition program was still converting below expectations and the bank needed faster, repeatable model-development workflows to turn transaction data into higher-quality leads and broader AI use cases.
Reported impact
The bank increased credit-card conversion, shortened model build and deployment cycles, and used the same IBM stack to put more than 15 analytics and machine learning models into production during 2024.
TPBank expanded the data platform it had already built on IBM Cloud Pak for Data by using IBM Watson Studio, IBM Watson Machine Learning, IBM Watson Pipelines, and IBM Analytics Engine to operationalize customer propensity modeling. The bank’s data team processed up to 25 GB of structured transactional data through a CRISP-DM workflow covering preparation, feature engineering, model development, and deployment, then used the resulting propensity models inside its digital banking channel to identify credit-card prospects while protecting credit quality. Cloud Pak for Data gives the team a shared foundation for building and deploying data-science assets, and the surrounding Watson components reduce the handoff and operational overhead required to move models into production. Over three years the same platform also helped TPBank build a broader data-fabric capability and scale analytics products across the bank rather than treating each model as a one-off effort.
TPBank ingests up to 25 GB of structured transactional data into Cloud Pak for Data, applies feature engineering and propensity modeling through Watson Studio, Watson Machine Learning, and Watson Pipelines, and then deploys the resulting scores into the bank’s digital channel to target likely credit-card applicants and support broader analytics products.
“15 data science and machine learning models developed and deployed with CP4D in 2024”
“24% conversion rate of credit card propensity model”
“Since 2021, Tien Phong Commercial Joint Stock Bank (TPBank) has been using IBM Cloud Pak for Data (CP4D) to modernize the bank’s data infrastructure.”
“To address this problem, the TPBank data team developed Credit Card Propensity models using IBM Watson Studio, IBM Watson Machine Learning and IBM Watson Pipelines.”
“the bank reduced the development and operation time of new models by 30% to 40%.”
Review how this reference was scored and trace each normalized claim back to its documented source.
Scoring methodology
Customer Reference Value measures usefulness as a customer reference. Evidence Confidence measures the quality and traceability of available evidence, not independent verification.
Traceability
Field coverage shows which facts each source supports. Dates are shown only when documented by the source.
Related
IBM Sample Apr 17, 2024 BS Brodnica BS Brodnica wanted to attract and onboard more commercial banking customers, but repetitive support questions and product guidance were consuming consultant and contact-center time and made it harder to deliver instant,… 1,000 conversations in 2 months Actionbot · IBM watsonx Assistant 
Google Cloud Sample Modified · Oct 9, 2026 Raylu Founded in 2022, Raylu is building the AI origination engine for private markets, replacing static tools and manual research with an always-on system designed for the speed of modern investing. Scaled infrastructure as Raylu grew from three to 30 employees Cloud SQL · Gemini platform · Google Cloud Secret Manager 
In-house Sample Aug 13, 2026 BBVA BBVA needed to improve customer service, streamline risk analysis, boost commercial productivity, and reduce administrative effort across commercial banking processes while giving bankers more personalized, faster… 86 percent of BEI agents in Spain AI Banker · Akira Project · AutoPF 