Intelligent document capture: how documents get in, and what to put where

Document capture is everything between a document arriving and that document sitting in the right place with the right data attached. It is the least glamorous part of document management and the part that decides whether the rest works. A document filed under the wrong type, split in the wrong place or indexed with a typo is, for practical purposes, lost.

This guide covers where documents come from, the three jobs capture has to do, how capture relates to the system that stores your documents, and how to choose between a capture station on a desk and a capture service on a server.

What intelligent document capture is

Older capture meant scanning: a person fed paper, typed a few index values and pressed save. Intelligent capture does the typing and the sorting. It decides where each document starts, what it is and which values describe it, and asks a person only when it is unsure. The word "intelligent" is earned by three behaviors: it handles layouts it was not configured for, it says when it is unsure, and it improves from corrections. Our intelligent document processing guide goes deeper on that reading step.

The four ways documents come in

Map every source before choosing a tool. Most organizations have more than they think: the scanner at the front desk, a shared AP mailbox, a folder a billing system writes to every night, and forms people fill in.

Two quieter sources are worth listing too. Office users save documents from Word, Excel or Outlook: Content Central's Office and Outlook add-in files them directly. And documents that already arrive as files from many senders can go to a cloud service such as Paige, which reads them and returns data by download, SFTP or webhook.

The three jobs: separate, recognize, index

1. 01

   Separate

   Decide where each document starts. The methods, from oldest to newest: a fixed page count (fine when every document is one page), separator sheets or barcodes between documents, and separation from the content itself. Our batch scanning and separation guide compares them in detail.
2. 02

   Recognize

   Turn the page into text (OCR), decode barcodes, and classify the document type. Image cleanup belongs here too: straightening, removing blank pages, fixing rotation. Recognition quality starts at the scanner, so scan at a sensible resolution and keep the glass clean.
3. 03

   Index

   Attach the values people will search and route by: vendor and invoice number, employee ID, loan number, claim number. Values can come from the page, a barcode, an email header, an XML file or a lookup against your own database. Unsure values wait for a person.

Click to enlarge

Capture front end vs document system

A capture front end gets documents in and turns them into files with data. A document system (document management, or a content repository) is where those files live: it controls who can see them, keeps versions, runs approvals and retention, and lets people search. The two are often sold together and often confused.

Keeping them distinct in your head pays off. Capture technology moves fast; your system of record should not. A front end that exports clean, named documents to a standard destination can be replaced without touching years of filed records.

How CapturePoint 6 and Content Central fit together

CapturePoint 6 is the high-volume scanning station. It splits stacks, recognizes document types, reads fields and line-item tables on the PC, checks line-item math, and puts anything doubtful in a review screen that says why it is waiting.

Content Central is document management on your own servers. CapturePoint hands it finished searchable PDFs with their field values over a dedicated API connection that an administrator can revoke at any time. CapturePoint can read a Content Central catalog to set up a scan job, or, in a new installation, propose and build the catalog for you. From there, Content Central takes over: full-text search, permissions, versions, workflows and approvals, retention.

Content Central also has capture of its own, which covers the sources a desktop station does not: mailbox capture, watched folders, XML hand-offs, capture forms, the Office and Outlook add-in, scanning from the browser, and printable QCard separator sheets. Anything captured without full index data waits in its Coding Queue for someone to complete, so nothing is filed half-described. The Content Central help article on how documents get in lists every path, and the email capture article walks through mailbox setup.

Desktop capture station or capture server?

Both are capture; they differ in who is present. A desktop station is a PC with a scanner and a person who scans and reviews. A capture server is a service that watches mailboxes and folders and works unattended.

Most organizations end up with both. In an Ademero setup, CapturePoint 6 stations handle the paper at the desks that receive it, and Content Central's background capture handles the mailboxes and system folders, with everything landing in the same library under the same document types.

Planning checklist

Work through this before buying or rebuilding a capture setup. It takes an afternoon and saves weeks.

• List every source

  Scanners, mailboxes, folders, forms, Office users. Note rough volume and who handles each today.

• List the document types

  Name each one and the fields people search or route by. Keep the field list short.

• Decide how stacks are separated

  Content-based splitting, separator sheets or barcodes, or one document per scan.

• Decide the naming and filing rule

  Folder and file names built from captured values, so nobody renames by hand.

• Decide what needs a person

  Which fields are required, and how sure the software must be before a value passes on its own.

• Pick the system of record

  Where documents live for their whole life, and who may see each type.

• Choose station, server or both

  Match each source to the table above.

• Pilot with real documents

  One department, a few weeks of real volume, the mis-splits and review time written down.

For accounts payable specifically, our AP document capture setup checklist goes step by step, and the CapturePoint help library covers job setup.

Common mistakes

• Capturing too many fields. Every field costs review time. Capture what someone will search, route or post by.
• Treating email as one source. An AP mailbox, a claims mailbox and a general inbox behave differently; give each its own rules.
• Leaving filing to people. If staff drag files into folders after capture, the naming will drift within a month. Let capture name and file.
• Ignoring the scanner. Recognition starts at the glass. Worn feed rollers, dirty glass and a resolution set too low cause errors no software setting can undo. Scan text documents in black and white or grayscale at a sensible resolution, turn on duplex where backs carry content, and clean the scanner on a schedule.
• Mixing document systems. If invoices land in one place, contracts in another and HR files in a third, people stop trusting search. Pick one system of record per kind of document and point every capture source at it.
• Skipping the pilot. A vendor demo on clean samples tells you little about your crooked, stapled, two-sided reality.

Questions