Bank turns to BPO for paperless loan processing
One of Australia’s major banks has eliminated paper-handling in the processing of loan applications by implementing a customised solution from business process outsourcer, Computershare Communication Services (Computershare).
According to the bank’s 2010 Annual Report, “investment in our back-office processing has yielded significant improvements in processing times, productivity levels and customer service rates.”
In addition to enabling a paperless workflow, “the solution has improved labour productivity by 200% and delivered a reduction of 20 days in the average loan settlement time.”
“Computershare was chosen to develop the loan processing platform after the bank investigated a range of different market solutions, however it could not get the required accuracy and had a large exception queue,” said Mark MacLeod, National Solutions Manager, Computershare Communication Services.
“We tried to buy a software package to fulfil our client’s needs and could not find one that provided the necessary accuracy for the 65% of ‘unstructured’ loan documents.
Computershare’s solution provides throughput of images with over 80% automated recognition for all document types.”
“Computershare’s solution solved the bank’s existing problem, and we are now receiving and defining large volumes of documents via fax and e-mail, swiftly capturing data, classifying and indexing documents on behalf of the Bank using a seamless automated process – saving time, educing error rates and costs.”
The output must conform to an interface specification to allow for automatic upload via dedicated links into the Bank’s workflow platform. All steps had to be performed in a highly secure environment within a 30 minute SLA from fax sending to output delivery. This allowed the Bank to provide loan approval within a 60 minute time frame, potentially removing a customer from the loan market at the point of sale.
Computershare’s existing data capture solution comprises both automated and manual processing components. To deliver on the massive volumes that were processed by the bank’s system every day, Computershare had to review and re-architect many areas and review the solution from ground up.
Computershare has two locations in Melbourne both hosting a scalable infrastructure environment to ensure business continuity. Load is shared between two sites and exceptions are distributed to operators located in production centres in NSW and Victoria.
Email is routed via a secure network link to Computershare’s redundant email gateway machines where the emails are scanned and delivered. Encryption is used throughout the processing stages to ensure that emails are encrypted at rest while residing on the email gateway machines.
Scanning of loan applications and supporting documents is managed by individual branches and brokers using existing Multi-Function Devices (MFDs), and then submitted to Computershare’s Capture Ingestion Service.
Computershare has integrated ‘best-of-breed’ OCR software and utilised its in-house development team to develop the auto-classification engine and production control system.
“Computershare has developed its own loan process automation solution to enable unstructured content to be accessed and processed for classification and data extraction with minimal exceptions, and therefore operator intervention” said MacLeod.
The solution is preconfigured to handle over 130 loan document categories and is able to automatically generate business rules based on learning sets.
It is currently processing over 120,000 document images per day for the Bank, at a throughput of 1400 images per FTE (Full Time Employee) hour.
“Removing paper from the loan process is incredibly complex. This client had to deal with large volumes of unstructured content and there were major change management issues to deal with” said MacLeod.
Computershare’s automated processing includes the following steps:
Document Normalisation – converts incoming files of various file formats into single-page image files that are suitable for further processing;
* Pre-process – image cleanup including despeckle, deskew, and automatic rotation;
* OCR – Optical Character Recognition to extract the textual content of images;
* Classification – Using the text content of an image, rules are executed against that text to classify the content to one of a pre-defined set of classifications;
* Extraction – Rules are executed on images of specific classifications to capture or extract values from the text content;
* Document Separation – Documents are received as a stream of images. Rules are executed against the text content of the images to determine document boundaries to identify the distinct documents within that stream of images;
* Page Numbering – A specialisation of the extraction function, specific rules to extract page numbers so that the images can be numbered and sequenced correctly; and
* Masking – edit or destroy specific areas of an image containing sensitive information, such as TFN numbers.
Exceptions and quarantine - Operator intervention where there are exceptions or certain steps are “quarantined” against automated processing.
Computershare developed the corporate imaging solution in a generic form so it can later support other business processes and divisions of the Bank as well as mortgage processing and is now marketing the platform in Australia and globally to organisations seeking an automated loan processing solution.