Speed Forms Processing With Automated Data Extraction

Financial Aid Independent Review (FAIR) was founded in 2001 to provide objective, comprehensive financial aid need analysis to private and parochial schools. The organization provides and processes forms that help schools best allocate financial aid funds to families that have the most financial need. FAIR doesn't make the allocation decision for the school; the company provides financial aid forms to the school, which are distributed to parents to complete (the form includes financial information and requires submission of proof of income). Parents mail the form to FAIR where it is processed, and FAIR provides the information to the school within three weeks of receiving the form.

FAIR processes approximately 30,000 financial aid forms (the organization has one form for private schools and one for parochial and has created 10 customized forms for schools that have requested such) for more than 445 schools each year, and 85% of the forms are processed between March and June. Until recently, this process was completed by manual data entry. When applications were received, the data from them was manually keyed into FAIR's proprietary software. In addition to its four full-time employees, FAIR needed to hire six to eight temporary employees during the peak months to complete an additional 1,700 to 2,000 hours of data entry. According to Dwayne Wolterstorff, executive VP at FAIR, the organization got to a point where they were having a lot of trouble finding reliable data entry help. "Even paying $15 per hour, it was difficult to find stable help," he says.

FAIR decided it was time to move to a data capture solution to automate the data entry process. It was understood when the company was founded that it would at some point move to document imaging. Therefore, the forms FAIR uses were created in text box format (so they could use a template-based imaging solution) to ease the transition from manual methods to electronic. Wolterstorff compared three vendors' solutions he had worked with in previous positions and chose ABBYY FormReader 6.5 data capture software, along with two Canon DR5010C scanners. "I chose ABBYY's product because I believed it was the most user-friendly, and because many other vendors use ABBYY's OCR (optimal character recognition) engine," says Wolterstorff. The solution would eliminate the need for data entry, because when a form is scanned on the Canon scanner, the FormReader 6.5 software can extract the necessary data based on a template. A template is created for each form, directing the software where to look for data on the form. The extracted data from the form is automatically sent to FAIR's SQL Server 2005 database.

To install the solution, Wolterstorff purchased two Dell computers to set up as scanning workstations. He then installed FormReader 6.5 on each of the computers and connected the Canon DR5010C scanners. ABBYY designed the first template, and Wolterstorff created the other 11 himself. During the creation of templates, only slight redesign of the forms was necessary since FAIR had planned ahead for scanning and automated data capture. Each template took about one day to create. "I purposely installed the solution during our slower period so I'd have ample time to create the templates, familiarize myself with the solution, and train the other employees before the busy season began," says Wolterstorff. He took two months to get used to the solution himself and write a training document before providing one-on-one training to the other employees.

Automated Data Extraction Reduces Labor Needs
Now, when forms arrive at FAIR, they are opened, sorted into bundles of 25, and placed in 'to be scanned' cabinets. This is done so that, at a glance, employees can easily count the number of forms needing to be scanned. Employees opening mail cross-stack income verification information that comes along with the forms (these documents are scanned in for electronic filing, but aren't read by the ABBYY software). The Canon scanner allows a bundle of 25 forms to be scanned in about 20 seconds. Once the forms are scanned, employees verify the data extracted by FormReader.  After the scanning and data verification process is complete, employees review the information against the income statements received with the forms. With the automation of data extraction, employees are able to process three times as many applications in one hour than they did with the manual data entry method. As a result, FAIR has been able to decrease the number of temporary workers it needs to hire from 1,700 to 2,000 hours to between 500 and 750 hours during its busy period.