Addressing Document Management Challenges

Credit unions everywhere face fundamental challenges with document-based information. Left unresolved, these significantly impact both compliance and competitiveness.

Regardless of their industry, employees who deal with document-based information understand how tough it is to access the right document at the right time, and ensure that documents used within a business process are complete and available in a timely manner.

Loan and Account Processing Issues

Credit unions are impacted more than many other organizations are due to the volume of documents required for engaging with existing and new members. Each product offered typically starts with an application and is supported by many other documents including, but not limited to: W2 statements, 1099 forms, paycheck stubs, bank statements and identification documents.

These documents are commonly supplied at various stages of the loan process. Many times, they are incomplete and/or incorrect. The result is that repeated requests are frequently made to the member after the staff has fully reviewed these documents. This causes unnecessary delays. Additionally, documents often come in through different channels, such as fax, mail, personal visit or mobile application. In these multi-channel cases, the status of documents may be unknown due to different employees involved with receiving and recording the documents. Unfortunately, this also causes delays and can even result in multiple requests to members for the same document.

Account Servicing Issues

From a servicing perspective, documentation problems rarely go away. The staff needs to reference documents, which can include member communication such as letters and emails, as well as lending documentation, invoices, contracts, agreements, work orders or proposals, signature cards, IDs, change of address forms and certificates of deposit.

Depending on document control processes and adherence to them, account documentation is typically not easily accessible in a way that provides an employee the ability to use this information during the member engagement process. For instance, many credit unions adopted document control systems in the past decade, but did not adequately develop a document taxonomy that suited their information access and management needs.

As a result, multiple documents are captured together and exist as a single file. Worse, this file may only have the member account number as metadata to search. Because of this, employees cannot easily locate this information in a timely manner to support a service engagement. Member representatives may find the necessary documents, but they have to review the entire package to find the information they need, which can be frustrating to both employees and members.

Identifying the Core Problem

Both of these problems stem from a common root cause: Lack of a proper document taxonomy and associated metadata to allow for more efficient and effective control over and access to account-based documents. While many credit unions understand the problem, they haven’t found a solution that really fits their needs. Either they establish and fund a project where staff members manually review each document, assign each to a particular document type, and then add the proper metadata, or they suffer through their chronic pain of service and processing issues. Most of time, the expense and resources required for consistent manual review and organization are too high.

Without a clear understanding of what documents are critical and the information contained within them, there is no way to improve the process.
Finding a Solution

There is no silver bullet to solving the problem many organizations large and small have identified. Taking the time and effort to understand what documents are used to support a process, and how they need to be described, is a fundamental need. “Assistive technologies” can make the process easier, less costly and more controlled.

For instance, clustering technologies allow a system to automatically group documents by likeness. This works in a similar way to a person separating a stack of documents and placing them into logical groupings. Without understanding what the documents are, humans can notice repeated similarities with documents and build different groupings of like documents. They do this by noticing the structure of the document, presence of visual elements, or specific words and phrases. The same can be done with technology that uses advanced text and visual analysis algorithms to group documents without any prior knowledge. This grouping can provide the basis for understanding what types of documents are stored within various repositories.

Moving forward from clustering is a related technology known as classification. It groups documents based on a particular rule set for each document class. The process can involve creating specific rules based on that presence of text or logos. Document classification can also involve more advanced machine-learning, where a sample of documents representing a particular document type is used to automatically “train” the system. The result is that rules are automatically created and improved upon over time.

Document classification can be done in a fraction of both the time and cost of the labor intensive alternative, leaving only those documents that are not successfully assigned to a class for manual review.

With classes defined and the required metadata identified, data extraction technology can be used to locate specific information that is used to support a particular class metadata. Using data extraction, applications and other forms are also evaluated to ensure all fields are populated, and those that can be cross-referenced with other account data can be validated automatically.

Summing Up

Through automation of document classification and data extraction, both business workflows and account servicing become much easier and more efficient. Workflows can be updated to include automated analysis to ensure all required documentation is present and in compliance, regardless of input channel and without manual review. Account servicing can include real-time access to required documents, and staff members won’t need to search through numerous files. This leads to faster processes and an improved member experience.

Greg Council is vice president of marketing and product management for Parascript. He can be reached at 303-381-3100 or greg.council@parascript.com.

Original Story