By Guest Author: Max Lien
Billions of pages move between businesses, their customers, and partners every year. Whether it’s mortgage applications, healthcare enrollments, or brokerage account opening forms, these documents contain mission-critical information that must be extracted and entered into the correct systems of record for downstream transactional processing.
To date, organizations have typically relied on a combination of manual data entry and outdated technology to classify and process these documents, resulting in overwhelmed systems and employees, slow processing times, high costs, and unhappy customers. Organizations spend roughly $60 billion each year on manual data entry, and according to McKinsey, across all occupations in the U.S., one-third of time spent in the workplace involves collecting and processing data.
Fortunately, solutions are available today to automate these processes. The explosion of Robotic Process Automation (RPA) and Machine Learning for the enterprise is changing the way organizations operate, reducing administrative burden and freeing up resources to focus on activities that drive a business forward. RPA works well for processes that are rules-based and involve structured data outputs (e.g., copying data from a spreadsheet and pasting it into the correct database). RPA solutions automate the highly repetitive tasks that hamper individual and organizational productivity, enforcing process and reducing the risks associated with human error. As a complement to these more rules-based approaches, Machine Learning models can perform specific tasks without using explicit instructions, relying on patterns and inference instead. These models then learn and re-train in response to the data to which they’re exposed.
While RPA and Machine Learning are each valuable on their own, together they’re synergistic. Not all inputs into a process today are neatly structured data, often making them inaccessible to RPA solutions. (In fact, research firm IDC estimates that 80% of data will be unstructured by 2025.) By integrating RPA with intelligent document processing solutions, enterprises can unlock and process data from a wide range of previously-inaccessible documents (PDFs, images, etc.). Structured data files can then be picked up by a digital workforce and automatically routed to the correct downstream systems for end-to-end automation.
Let’s consider what this looks like in the real world. Every day, Accounts Payable departments receive invoices from various vendors and suppliers, both electronically and physically. These invoices may need to be scanned and organized before key information, such as invoice number or total amount due, can be plugged into the system of record for payment. Leveraging the latest in Machine Learning, intelligent document processing solutions can classify the pages as an invoice, locate the relevant fields, and automatically extract the information with exceptional accuracy. This is particularly challenging – and valuable - because there is no standard invoice format across vendors/suppliers, and relevant fields for extraction can be located in different places across documents.
Whereas before one large American insurance company relied entirely on manual invoice processing, which often meant missed SLAs, using the power of RPA and intelligent document processing, the insurer is able to move multi-page, hand-annotated and scanned Accounts Payable invoices from a centralized inbox into an intelligent document processing platform for automated classification and data extraction. Structured data output files are then picked up and routed to the correct downstream systems, improving throughput, reducing error rates, and freeing up valuable employee time.
Machine Learning can take documents and turn them into data that is immediately useful and actionable by RPA. Coupled together, these two technologies have the ability to improve business processes, save time, and reduce costs, ultimately enhancing the overall customer/stakeholder experience.
Max Lien leads strategic partnerships at HyperScience, the intelligent automation company that enables data to flow within and between the world's leading financial services, insurance, healthcare and government organizations, including TD Ameritrade, QBE and Voya Financial, among others. Customers who use HyperScience experience as much as a 67% decrease in error rate, a 10-fold increase in processing capacity, and a six-hour reduction in meeting Service Level Agreements (SLAs).