References /

SAS Institute Inc.
Tel: +44 (0) 1628 490490
For more information on SAS, please visit www.sas.com/uk

References

Printer-FriendlyPrinter-FriendlyPDF PDF

UK References

 

SAS® High-Performance Analytics underpins growth strategy at one of the UK's largest banks

A new approach delivered a fundamental rethink of the way customer data was managed and processed, allowing the bank to effectively handle big data volumes, and to gain more flexibility through improved pricing insights.

One of the UK's largest banking organisations, with tens of millions of customers, needed a new approach to analytics to underpin the top priority of its board. The new approach delivered a fundamental rethink of the way customer data was managed and processed, allowing the bank to effectively handle big data volumes, and to gain more flexibility through improved pricing insights. By deploying SAS® Grid Manager, SAS® Scoring Accelerator for Teradata and SAS® Analytics Accelerator for Teradata over 176 cores, the bank now has a High-Performance Analytics (HPA) suite able to refresh individual customer models for intraday pricing, which is worth tens of millions to the bottom line.

The strategy set by the board will require investment of £2 billion over three years to revitalise key brands, triple the number of high net worth customers in the sales cycle and increase average income per customer. It quickly became clear that the board's objectives couldn't be achieved without significant change to the way the company understood and interacted with its customers.

The company's approach to analytics meant it took months for models to reach production and consequently forecasts were out of date as soon as they were produced. This had a direct impact on customers as mortgage and loan pricing could only be carried out on a monthly basis. Even more challenging for the marketing team were limitations meaning pricing flexibility was limited to customers approaching the end of their loan. There was no way to offer a new price to a customer who was considering moving provider part-way through their mortgage.

Like many in its industry, the bank was limited to modelling just 20% of its customers due to the sheer size of its data. The process of taking a model from conception to execution was complex, including significant manual intervention such as the need to write SQL code within the Teradata warehouse environment. Data needed to be extracted, transformed and loaded (ETL) into the bank's existing Base SAS® environment which ran on PCs. Processing limitations meant data had to be heavily sampled, reducing effectiveness. No direct link between Base SAS and Teradata existed, meaning each model needed to be manually recoded before deployment.

Building an integrated HPA framework
With the introduction of SAS Grid Manager, SAS Scoring Accelerator for Teradata and SAS Analytics Accelerator for Teradata over 176 cores, the bank's analytics process has been transformed in terms of its underlying computing infrastructure, performance, and the advanced 'next best action', 'pricing' and 'propensity' modelling that is now possible.

Laying the foundation with grid computing
Moving to a grid computing model was necessary in order to remove processing constraints, to reduce the cost of underlying infrastructure and to handle huge data sets. One hundred and seventy- six SAS Grid Manager cores were deployed, enabling high numbers of analytics and data integration jobs to be managed and balanced whilst running across 11 HP DL 465 blade servers, each with 16 cores. The move opened the door to commodity computing and the grid proved to be ten times more cost effective than the alternative IBM P Series hardware proposition. A high volume of big data jobs can now be run across multiple cores in parallel, taking just minutes and drastically improving time to action.

SAS and Teradata combine for in-database delivery
The bank's marketing team realised that bringing models to fruition rapidly was key to delivering on the board's objectives. Following implementation of SAS® Enterprise Miner™, the bank is now able to construct models rapidly with a Graphical Interface, sharing developments with peers and removing the coding task that was previously undertaken in Base SAS. Users can now prepare data for models within the newly installed SAS® Data Integration Studio rather than extensively querying records within Teradata, a process that required a six month SQL training cycle.

According to the Principal Technical Account Manager at SAS, "We have established an end-to-end process where the model moves seamlessly between SAS environments. There's no recoding, no need for checking, and execution uses the parallel environment within Teradata. Big data models can be live within days."

When a model is built in the new SAS® High-Performance Analytics environment it can be run directly within Teradata processes using SAS Scoring Accelerator for Teradata, meaning that even models relating to huge data sets of 40+ million customers can be run within minutes. The powerful HPA framework means the marketing team is now able to deliver individually priced loan offers to customers at any point during the term of the loan.

The introduction of HPA means the bank is examining integration of unstructured data, such as customer calls, in order to model the propensity of individuals to churn. HPA has opened the door to applying such models on a daily basis so that 'next best action' insights can meet customers' unique needs. For example, the bank now understands 'the next best action' it should recommend to an individual as it relates to their unique situation, whether it is a new offer or a proactive service call.

SAS Client Director describes the business impact of HPA: "Daily pricing is only possible using High-Performance Analytics that negate the challenges posed by moving, transforming and preparing big data sets across environments. The business case for daily pricing of mortgages alone is worth many tens of millions to the bank."

HPA delivers on rigorous requirements
SAS® technology was already used across other areas of the institution including risk; however, as with any critical project, the bank insisted on a robust Proof of Concept. SAS Scoring Accelerator for Teradata and in-database processing were tested against 50+ million records using one of the marketing team's regular models with no adverse effects on the Teradata environment, a CPU efficiency rating over 90% and all queries ran in parallel as planned. SAS Analytics Accelerator was also tested on circa 1 million records with no adverse effects. Each Proof of Concept run was completed inside 60 seconds within the Teradata test server.

Copyright © SAS Institute Inc. All Rights Reserved.

Large UK Bank

Business Issue:
Requirement for a High-Performance Analytics framework to refresh individual customer models for intraday pricing and propensity analysis to deliver a 'next best action' understanding of customer needs. HPA is essential to delivering the board's strategic objectives.
Solution:
Integrated SAS® High-Performance Analytics framework comprising: SAS® Scoring Accelerator for Teradata; Base SAS®; SAS® Analytics Accelerator for Teradata; SAS® Enterprise BI Server; SAS® Enterprise Data Integration Server; SAS® Enterprise Miner™; SAS® Enterprise Model Management; SAS® Forecast Server; SAS/ACCESS® Interface to PC Files; SAS/ACCESS® Interface to Teradata; SAS/CONNECT®; SAS/ SECURE™; SAS/STAT®.
Benefits:
Drastically improved time to deployment for customer modelling from months to days; ability to deliver daily pricing for mortgage and loan products as they relate to individual customers; grid computing infrastructure which is ten times more cost effective than the previous system; ability to understand customer needs in terms of 'next best action'.

Daily pricing is only possible using High Performance Analytics that negate the challenges posed by moving, transforming and preparing big data sets across environments. The business case for daily pricing of mortgages alone is worth many tens of millions to the bank.

SAS Client Director

SAS UK

Read more: