Distribute Imaging To Transform Paper Processes

by Dan Bolita

Time is money, as the cliche has it, and this simple fact continues to fuel interest in distributed document imaging. For the lion's share of business processes that still involve paper documents, distributed imaging tools and techniques allow organizations to speed processing and cut the cost of transactions initiated across a region, across the country or even around the globe.

In the past, distributed imaging systems were custom built at great expense, and they tended to rely on expensive dedicated lines or cumbersome FTP processes. With the ascendance of the Internet, however, more and more off-the-shelf products are available to easily support scanning or indexing, data entry and validation in remote locations.

Distributed scanning systems allow dispersed organizations to collect document images from multiple customers, remote offices or satellite operations. Relatively small volumes might be collected from scores or even hundreds of sites, or larger volumes might be captured and transferred from a handful of production-level sites. Either way, this approach eliminates shipping costs and delays, cuts operating costs, and speeds transactions for faster customer service and access to revenue.

Systems that support distributed indexing, data entry and validation allow you to move these labor-intensive steps out to home workers or remote sites in low-cost labor markets. With larger volumes, the work might be outsourced to service bureaus of offshore data entry operations. This approach is more common in data-intensive forms processing applications. The larger the volumes, the more crucial reliable image conveyance, tracking and monitoring features become.

The Poster Child for Distribution

There is perhaps no better example of a business that can benefit from distributed imaging than an insurance company. The typical insurance company has hundreds or even thousands of agents in the field who must submit both policy applications and claims. These documents tend to be long, complex and data-rich, and many of these forms must be retained in long-term archives.

The research firm Sherwood International, London, suggests that Internet technology could help insurers cut costs by 60 percent, mainly by reducing the costs of re-keying data and other fragmented manual processes. While some insurance processes are moving online, the industry as a whole has not yet capitalized on the advantages of electronic delivery, according to Meridien Research, Newton, MA. The researchers attribute this to "channel conflict," primarily with traditional agent distributors, and to the complexity of many insurance products.

"Although much attention has been focused on Web-based policy origination, online policy administration is the area where we believe the greatest potential lies for the insurance industry," Meridien reports. "We expect the origination process for life insurance policies to move online only gradually."

While insurance companies may still be originating policies on paper, that doesn't mean they can't use the Internet to take the transaction from there.

ING ReliaStar, the Atlanta-based unit of Dutch financial services giant ING, is doing just that with the help of integrator EDS, Plano, TX, and capture vendor Captovation, Minneapolis.

ING ReliaStar exemplifies the use of a hybrid class of remote/centralized capture systems. The company's documents are collected in a number of formats at a central location in Minot, ND. These documents include life insurance applications from field agents as well as physician statements, motor vehicle records and other support documents. The documents are scanned, indexed and sent to a batch server in Minneapolis over a dedicated T3 connection. This link handles telecommunications in addition to network traffic. The remote site is scanning 75,000 to 100,000 images daily.

"We're looking to do scanning and indexing at three locations," says Celia Weissman, senior programmer analyst for ING ReliaStar. The goal, she says, is to minimize document handling. "The idea for us is to process each image once; if a separate indexing process were required, we'd have to touch them twice. Now when we're done [capturing] images, they're automatically indexed as well."

In addition to combining capture and indexing in a single-step workflow, ING ReliaStar is establishing multiple remote locations where images can be received. Weissman points out that this will allow the company to share the load between its capture sites. "We can transmit batches to other locations to accommodate the peaks and valleys," she explains.

Faster image capture also hastens data export to line-of-business applications. The project will also enable ING ReliaStar to more easily share information with re-insurers and other business partners.

"As it is now, we fax or send FedEx," Weissman explains. "In a paper environment, we had a procedural workflow; in an imaging environment, strategic workflows are being built."

ING ReliaStar is just beginning to plan for a Web-based system. Weissman says she's well aware of potential process improvements that might come from a Web-enabled system. "A number of our larger agencies are looking at it," she says. "We just can't support that yet."

Weissman agrees with Ken Peterka, president of Captovation, when he says that there are still technical and regulatory obstacles to Web-enabling the insurance industry. "When a Web-based form is filled out, legally, this needs to be [preserved as] a document," Peterka says. "This is a huge issue for processes that require records retention."

Adopting Browser-Based Scanning

While ING ReliaStar's implementation involved just a few sites scanning in high volumes, the more common distributed scanning scenario involves large numbers of sites scanning in low-volumes (say a handful of policies and claims per day). In the former scenario, ensuring adequate bandwidth is the biggest concern (an issue ING ReliaStar resolved with a dedicated, high-bandwidth network connection). In the latter, low-volume scenario, ease of installation, administration and consistent document control are the biggest challenges.

"For lower-volume distributed capture, you can either scan and index batches at a remote site or just scan the batches and index them at the central site," says Peterka. "Either way, you need to have control over how they're scanning and/or indexing different types of documents. We've tried to turn it into a copierlike walk-up application with the help of browser-based scanning."

Captovation's ecNet distributed scanning system employs a central server and easily downloaded Active X plug-in to ordinary Internet Explorer browsers. The system supports the TWAIN drivers typically used with low-volume scanners, but with a high-bandwidth DSL connection, Peterka says the system can transfer a typical batch of 100 50K images in about three minutes (see table below). Images are cached on the local client until the server software confirms successful receipt of the batch at the central site. (To meet production-level demands, Captovation deploys conventional Windows-based scanning and indexing modules from its eCapture suite.)

How Fast Can You Transfer Images Via the Internet? Listed below are estimates of the time it would take to transfer a batch of 100 50K images using common Internet connections. These conservative estimates add 10 percent to estimated transfer times to account for overhead and true throughput (e.g., 56K modems typically communicate at less than 52,000 bits per second).

The real beauty of browser-based scanning is that you won't face software installs at the many distributed scanning sites you plan to deploy. Form and document workflows created at the central site and hosted on the ecNet Server are immediately accessible to all users via a Web browser. Changes in workflows and upgrades take place transparently to the end user. The interface can be made as simple as possible, with drop-down menus offering a choice of common document types with pre-defined scan settings, index schemes and workflows. Barcodes can also be used to automate indexing and processing workflows.

"The concept of scanning through a Web browser is new," says Peterka. "It has taken a while to get our resellers to really understand it and feel comfortable with it."

Captovation isn't the only company pursuing this approach. Prevalent Software, for one, has a similar, browser-based TWAIN scanning client available with its Quillix capture software. The Colorado Springs, CO, vendor has installed the software for customers including Protective Life, Birmingham, AL, an insurer that is using the software to capture documents sent in by independent agents.

In April, Kofax Image Products, Irvine, CA, lent heavyweight support to the browser-based approach by adding a Web indexing/validation module to Ascent Capture 5.0, the latest release of the company's document imaging software.

The Ascent Capture Web Validation Server breaks down batches into individual images and queues them for download to individual operators. User profiles manage which images can be viewed and worked on by which users. Full-size TIFFs are cached in RAM on the local machine one at a time for OCR/ICR validation or key-from-image data entry. As the user completes each data set, the corrected or newly entered data is sent back to the server while the reference image is wiped from local memory. In the meantime, the next image has downloaded and is ready to be worked on.

According to David Oldfield, Kofax's senior product manager for Ascent Capture, browser-based users can be supported with a connection as slow as a 56K dial-up. The browser supports the same keyboard shortcuts - field tabbing, rotation, zooming, image reject, etc. - used with conventional clients. However, database lookups and scripting aren't supported in this release.

"It's our expectation that all capture functions could be performed through browsers," says Oldfield. "It's perfectly reasonable to expect this in a LAN or WAN environment, but it's less clear how an organization might do it over the Internet. Realistically, you'd need at least a DSL connection to do a lookup, bring the values back and have it happen rapidly enough that the operator wouldn't be sitting there waiting. Another possibility would be mirroring data at remote locations so you can get away with lower-speed connections."

John Stetak, vice president of marketing at ActionPoint, San Jose, CA, contends that most of the current browser-based tools are suitable only for ad hoc use. "There is a place for low-volume tools, but distributed capture means a lot more than scanning through a browser," he says. "If you're dealing with production-level work and you're outsourcing or sending images overseas, you're going to need robust server-based software to move batches from center to center without losing data or images. You're also going to have to support validation and key-from-image processes with pre-fetching, caching and load balancing to support high-speed data entry."

Stetak contends that ActionPoint's client/server document and data capture modules beat peer-to-peer software in supporting production-level scanning and processing. The company uses a server-based IACopy module to ensure batch integrity as images are passed from remote scan stations to central servers or onto remote indexing, validation or data completion modules.

Outsourcing Data Entry

Many would agree that the tougher, costlier stage of document imaging is the indexing, data entry and validation performed once the scanning is done. This is certainly the case when the documents in question are complex forms requiring lots of data extraction. Here, too, organizations are turning to distributed approaches to outsource work to home workers or service centers in low-cost labor markets.

Microsystems Technology Inc., Tampa, FL, was an early pioneer of distributed processing technology. The company offers a Java-based client that supports many of the same data formatting and validation routines built into its conventional OCR for Forms software. The distributed client also manages the process of downloading images and synchronizing completed data with the central server.

Captiva Software, San Diego, is among the latest forms processing software vendors to add distributed processing options. The company's recent release of its 4.0 FormWare system offers both distributed scanning and validation/completion options. Capture Resource, a Bristol, PA-based service bureau, says these features will help it meet demands for offsite data entry.

"One of our clients is a major bank lockbox provider that wants to keep the [checks and remittances] in Washington while utilizing our keying expertise in Bristol," says Sean Sliwinski, Capture Resource's vice president of technology. "The distributed processing capabilities will allow us to provide this service while offering the firewall protection that the bank requires."

For distributed scanning, Captiva deploys a conventional Windows-based scan client at remote locations, but it adds scheduling and synchronization software that periodically uploads cached images to a central Internet FormWare Server. The Completion client is also Windows-based, but it can be downloaded over the Web along with image batches and associated rules for data validation. Workers complete data entry or OCR/ICR validation work offline and resynchronized with the central Internet FormWare Server whenever they choose.

"Many of our customers already support distributed operations with virtual private network connections, but state tax agencies and other customers have asked for a way to support at-home workers using low-cost Internet connections," says Chris Gollery, Captiva's FormWare product manager.

Other forms processing vendors that have introduced support for remote operations include Cardiff Software, Vista, CA, which has introduced a distributed data entry/validation option. ReadSoft, San Diego, has introduced distributed scanning options. Top Image Systems, Carlsbad, CA, supports both remote scanning and validation through its Integra Enterprise Controller.

Distributed imaging has long been possible through custom integration or the use of fax machines, but the former route is costly and time consuming while the latter presents low-quality images that are difficult to automate. The growing list of out-of-the-box tools will make it easier to set up more efficient, cost-effective approaches to document imaging. Add to that mix easy-to-use walk-up devices such as HP Sender-type devices and networked digital copiers, and you have the makings of a new era in transforming paperwork into speedy electronic transactions.

Dan Bolita (danboli@midcoast.com) is a contributing editor to Transform Magazine and editor of Today, the journal of The Association for Work Process Improvement, in Boston. Additional reporting by Doug Henschen and Adam Throne.