Delving into document imaging

Dan Deserto: I think it's really important to understand this distinction. "Imaging" is a model or process of storing and managing your documents, using a software program. "Scanning" is simply the physical process of running a paper document through a scanner-and is just one way to put documents into your imaging system.

Steve Anderson: Say my agency has decided to make the move to an imaging system. What's the first thing I need to do?

Dan Deserto: Your first important decision is to choose what processing model you'll use. There are three basic models.
?With a pre-process, or mailroom, approach, all paper documents are scanned into your system at one central location, such as the mailroom. The scanned images are then sent to the appropriate people in the agency. ?With a local, or desktop, model, paper documents are distributed to everyone in the agency, and employees scan their own documents into the central system.
?With a late-capture, or post-processing, model, paper documents are distributed throughout the agency. Employees then process the information from those documents, such as by entering an activity in the agency management system using information from a paper document. The paper documents are then sent back to a central location for scanning into the system.

Dave Lunceford: One of the big advantages of using the front-end capture method is that you have what I call a "single point of procedure." If the scanning is being done at every desktop, you have a large number of people deciding what to scan and keep, versus what to throw out. If the scanning is being done at one location before distribution, those decisions are being made more consistently. Of course, this means that your scanning person-perhaps a mailroom clerk-would need additional training to be able to decide what gets scanned and what doesn't.

A possible drawback to the early-capture model is that account reps and CSRs will be using scanned images of documents for much of their work. Some people might rather have the paper in front of them as they enter information into an agency management system, instead of using the "alt-tab" or having two windows open on the screen at once. This is something to consider. Possible solutions include using a dual-monitor system on each desktop or purchasing a large monitor. But that is a discussion for another time.

Ted Baker: In the past, there was a fairly significant objection to the local (desktop) scanning model. Account reps and CSRs are some of the more important employees in an agency, and agencies didn't want those employees spending all their time running paper through a scanner. What has changed to make a local scanning model more feasible is that agencies are receiving less paper to begin with. More and more carriers are sending documents to the agencies electronically. So if you want your CSRs to do the scanning, they have less scanning to do. The advantage of this model is that CSRs and account reps likely have more expertise than a mailroom clerk or other employee in deciding what needs to be scanned into the system and what can be thrown out.

Steven Finch: The late-capture model combines advantages from the first two. Like the desktop processing model, it allows the CSRs or account reps to decide which documents should be scanned and which should not. Another advantage is that many employees want to have the paper document in front of them as they process the information, particularly if it's a multiple-page document. For example, a CSR using an eight-page document to create an activity in an agency management system would rather flip between eight pages of a paper document than have to navigate among the pages of a scanned document.

The increasing number of documents coming from carriers electronically is not a problem with this method, but it's something to be aware of if you're using a late-capture process. If your workflow model has CSRs sending paper documents back to the mailroom for scanning and indexing in the system, you need to account for electronically received documents that also need to be saved.

Dan Deserto: Sometimes it makes sense to consider a "hybrid" model as your fourth option. For example, an agency might decide that the late-capture method is their most efficient choice. The CSRs receive paper documents, process the information and send all documents back to the mailroom for scanning. The agency principal, however, might want to keep certain documents "closer to the vest," and thus might choose to use the local scanning method himself. If the agency has small branch offices in a rural area, it might be most efficient to scan documents for those offices at the central location and send them electronically. In this case, an agency might be using a hybrid of all three models.

Steve Anderson: I've decided what processing model I'm going to use, and it's time to buy a scanner. What should I be looking for, and how much should I expect to pay?

Ted Baker: Starting with the small desktop models, a Visioneer XP 200 scanner is about $250. It's about the size of a three-hole punch, and it can survive heavy use for four or five years. The next step up is the Visioneer XP 450, for $600. This version has an autosheet feeder and scans about 20 pages per minute. Fujitsu has a duplex scanner for around $800. This scanner will scan both sides of a two-sided document at the same time.

Dave Lunceford: Slightly more expensive, the major brands are Canon, Fujitsu, Ricoh and Panasonic. These are all good scanners, with price varying according to the page-per-minute speed. I've seen full-color, duplex scanners that can scan 160 pages per minute, and they sell for $15,000-far beyond what an insurance agency needs. A speed of 20 ppm with one of these scanners should be around $1,000.

An important consideration for an agency is how easily a scanner part can be cleaned or replaced. The two parts most likely to need maintenance are the rollers that pull paper through, which will eventually need replacing, and the scan head, which must be cleaned occasionally. Is this something an agency can do itself, quickly and easily, or does a technician need to come out and do it? Find out before you buy.

Steven Finch: If you get a duplex scanner, make sure your combination of hardware and software has "blank page detection." Some insurance policies are printed on both sides for the first 10 pages of a document, and have the second side blank for the next 10 pages. If your scanner can detect this and skip the blank sides, you save time and electronic storage space.

You might also want to spend a little more to ensure that your scanning process incorporates file compression, to reduce the size of scanned files. Some people have said this is not important, since hard-drive storage space is so cheap. But that's not the issue. If your scanned files are too large, they're more difficult to work with, and some carriers may not accept your electronic submissions.

Dan Deserto: Some clients we work with have asked about the all-in-one combo machines that can do copying, scanning and faxing. I always say that sure, these work, just like putting your laundry in a dishwasher works. The clothes will be cleaned, but not as well as if you put them in a washing machine. The same is true with copier/scanners. Canon makes such machines, for instance, but Canon itself says that the image coming out of the machine is not as clean as it would be with a scanner that does nothing else.

A relatively new feature that has become available on scanners is foreign object detection. A lot of documents your agency scans might be bound together with paper clips or staples, and forgetting to remove some of these could damage the scanner. The Canon scanner will notice this, momentarily pause the scanning while you remove the object before it goes through, and resume scanning.

Steve Anderson: As I begin implementing scanning and imaging in my agency, should I start slowly or put a system in place throughout the agency, all at once?

Dan Deserto: I prefer to limit the "first pass" of implementation to small, controlled groups. We limit agencies to no more than two departments and 20 total employees for this first pass. We do this to ensure that the implementation goes smoothly and that the rest of the agency sees how well it works.

Steven Finch: We take a different approach on the same principle. We think the success of an implementation should be visible right away, to help the entire agency buy into the new system. So we like to do the implementation on a large scale from the start. But before we do this, we conduct a "pre-implementation." We get several key employees from the agency, lock them in a room for a few days, and have them think through basic issues such as the naming system they'll use for electronic files and what they'll either put in their system or throw away. The chances for a total agency "mental buy-in" and a successful implementation increase when the agency has taken control of the process from the start and personalized it.

Dave Lunceford: I agree with both of these approaches. Agencies that have several small, remote locations should definitely consider a step-by-step approach. It might be too difficult to get people from every location together at the same time and place for an implementation. So the best choice may be to do the "beta" testing at the central location. When everything is in place and running smoothly, you can expand it one location at a time to the branches.

Ted Baker: The size of the agency should be a factor in your approach. If the agency has 75 or fewer employees, for instance, you could decide to bring everyone on board at once. But if you have 150 or 300 employees, that might be too difficult to do even if you have just one location. And the key is to decide not just how to implement an electronic filing system, but to decide what paper you will keep and what you will throw out.

Steve Anderson: What about information I already have stored in my agency management system, perhaps as file attachments? How should I move that to my imaging system?

Dan Deserto: Since you're still going to be using your agency management system, one approach is the "day forward" approach. You start using your new imaging system on a certain day, and all digitized documents are stored in your new system from that day forward. Images attached to agency management system files before that date are kept in your agency management system. The other choice is obviously to extract data from your agency management system and put it into your new system. Your options for doing this vary by which management system you're using, and you may run into technical problems along the way.

Steven Finch: I agree that a "day-forward" approach might be best. If you've decided to use a true document management system, rather than just attach files in your agency management system, that means you have decided that your current system's file-attachment capability doesn't completely work for you. If you don't like it, you probably wouldn't want to re-create that data in a new system. I think putting this old data in a new document management system compromises its integrity and "buy-in."

Steve Anderson: How do you keep track of all the documents you're scanning into a system?

Dan Deserto: Most document management systems get indexing information from an agency management system and can associate bar codes with indexing information in a correlation table. In other words, a table is created, using information from your agency management system, that contains identifying information about clients' folders. By scanning a bar code along with a document, you can then electronically file the document in the correct place, without having to navigate through the system "manually."

Dave Lunceford: While creating bar codes was complex in the past, today they are nothing more than a font to be installed in your system.

Steve Anderson: What's the role of optical character recognition (OCR) in this process?

Dan Deserto: I don't think there's widespread agreement on the use of OCR, which is a system that reads the characters on a page. In my opinion, the technology has gotten much better recently and does have a use within document management systems. That use is not the indexing of documents themselves, but perhaps the ability to find specific text within a set of indexed documents.

The problem with OCR is that even today it is considered only 97% accurate at best. So you're OK if you're using OCR to read document names within your indexing system, as long as you can afford to miss 3% of your documents. But suppose you are looking for a report about a specific topic within a series of documents. Doing a full-text search would be easier with OCR.

Dave Lunceford: I agree that you should not use an OCR system to read information off a document for the purpose of naming and indexing the document. Even the 97% claim about OCR accuracy should be looked at carefully. What that means is that an OCR system scans each character and recognizes the character correctly 97% of the time. That's each character, not each word. So if you multiply that through a 20-character word, the word is recognized correctly only 56% of the time. That's why I've said for years that OCR is great for non-mission-critical documents. Meaning, it's great if you really don't care whether you can find a document again.