Headquartered in Virginia with roots that trace back to 1902, our client is a publicly traded banking corporation with 230+ branches and ATMs located throughout Virginia, Maryland, and North Carolina. The bank reported revenue exceeding $718 million in 2021, offering customers a range of financial solutions including checking, savings, small business, and commercial business bank accounts; mobile banking; credit cards; mortgages; lines of credit; home equity, car, and personal loans; and more. It also offers wealth management services like financial and retirement planning.
The post office delivers about 100 mail forms to the bank every day for returned mail with undeliverable addresses, whether the customer moved or the address was incorrect. Most times, the form is stamped with a forwarding address that must be updated in the client’s core Horizon banking platform.
Manually marking customer profiles with a “Return Mail” code in the Horizon system or updating them with the new address was a time-consuming and tedious process, consuming about 1,000 employee hours per year.
Changing the address required an employee to upload a picture of the mail form, validate the new address, and key the new address into Horizon. Such mundane manual work is prone to human error, forcing employees to rework about 2% of transactions.
The client knew intelligent automation could increase the speed and accuracy of its process for resolving high monthly volumes of inaccurate customer addresses.
SOLUTION & APPROACH
Auxis was hired to assess, customize, and support the best intelligent automation solution for the client’s needs.
The Auxis team started with a feasibility analysis that tested results delivered by four different OCR (Optical Character Recognition) technologies. Essentially, OCR solutions automate the process of converting unstructured formats like the client’s mail form images into machine-readable, searchable text.
After running 50 sample transactions on each technology, it was determined that UiPath’s Document Understanding tool delivered the most accurate OCR results. Auxis also educated the client about future benefits that could be achieved by applying Document Understanding to more advanced bank processes once it owns the license.
One of the latest innovations by market-leading Robotic Process Automation (RPA) platform UiPath, the Document Understanding tool eliminates manual extraction headaches by using AI-enhanced robots to read documents of any difficulty, extract desired data, and act on the information with a high level of accuracy and reliability. It can automatically determine the location of document data if the document type or format changes, or if it’s rotated or skewed.
Machine Learning also delivers continuous improvement, increasing bot training and accuracy every time a process is performed.
After understanding its benefits, the client chose to install the Document Understanding tool, despite higher licensing costs than other options. The Auxis team customized the UiPath solution to meet the client’s specific needs and continues to provide maintenance and support.
InCustomized steps include:
After returned mail forms are scanned into the bank’s imaging system, a bank employee manually adds the forwarding address and tax ID number to the file since the mail form’s post office stamp and handwritten Tax ID identifying the customer are often difficult to read. The employee also identifies the scenario as a Hold Mail “yes,” “no,” or “add” - triggering different rules for the bot.
The bot reads and interprets the PDF file, “cleaning up” extraneous markings to locate address information.
The bot validates the Tax ID number and customer information.
The bot disables the incorrect address on the customer’s Horizon profile so mail is stopped and looks for an alternative.
The bot assesses whether forwarding addresses supplied by the post office are valid U.S. addresses. It also recognizes if the information represents a physical street address or a P.O. box, which are recorded as alternate addresses.
If an address is deemed invalid or scan quality renders the mail form unreadable, the bot sends the transaction to a human employee to review. Transactions are also sent for review if the bot can’t find a replacement for an incorrect address.
Validated addresses are updated on the customer’s Horizon profile by the bot.
The Auxis team regularly adds intelligence that helps the bot recognize valid address components that weren’t part of its original programming, such as unusual abbreviations or spellings.
The Document Understanding tool successfully uses a rules-based approach to intelligently automate processing of the high volume of returned mail forms for the client.
Key results include:
84% Processing Accuracy
Document Understanding weeds through extraneous markings on the mail form to pick the address correctly with 84% accuracy, relying on rules and parameters that help it detect valid address components like state, street numbers, and zip codes.
Human Staff Shifted to Higher-Value Work
A single robot performs the work that used to consume two bank employees, averaging 3-6 minutes per transaction based on complexity. Human staff now only spend about 200 hours per year on returned mail forms, mostly investigating complex process exceptions, saving them from the operational mundanity of processing returned mail and letting them focus on higher-value tasks and objectives.
Bots automatically generate activity logs, enhancing visibility into the process and creating an easy-to-follow audit trail.
Bots don’t make mistakes, eliminating the risk of human error.
Pleased with its automation results, the client has started exploring additional uses for the Document Understanding tool on more advanced processes with Auxis.