Best Practices and Additional Tips

1.     Use electronically produced pdf images whenever possible
The success of OCR technology depends on the quality of the image. As noted above, the best results are obtained when you use pdfs that are electronically produced (those emailed to you or downloaded from a vendor’s site). When invoices are scanned, the scanning process often degrades the quality of the image.
2.     When scanning, be sure to scan in black and white at 300 dpi.
3.     Verify System Settings in TimberScan admin:
     • Invoice Recognition Percent – set to 50
     • Image Resolutions for Capture and Final – set to 300 dpi
     • Remove Leading Zeroes from Invoice Number – check if your vendors use leading zeroes on invoice numbers and you do not want them to prefill on Capture invoices
     • Accounting Date Usage – select an option to default on Capture invoices
     • Default Invoice Date Format – select a format to default on Capture invoices
4.     Creating Capture templates is a learning process. Keep the number of templates small and manageable; test out each template to ensure Capture can read it effectively. Before creating a large number of profiles for a vendor or a large number of templates, consider limiting the number of profiles (and/or templates) and testing each for success. Spend your time learning to be an efficient template producer.
TIP: There is a “trade-off” with creating templates and profiles. Do you want to spend your time creating, for example, 50 profiles for one vendor or create multiple templates for many vendors?
5.     In the early learning stage, you may choose to ‘Bypass Failed Doc’ queue if many images are going unrecognized and you need to code the invoices for export to Timberline. Later on you can modify the template to have any rejected/unrecognized invoices flow to the Unrecognized Capture Documents queue.
    Look for instructions and messages as well as your results in the left panel of the create templates window.
    Keep the box size you draw on a template as tight as possible. Capture’s pdf text layering reads the image multiple times and can pick up ‘extraneous’ markings which may cause the template to go unrecognized.
    The default for reading text in a box – ‘Closest to the center of the box ‘– usually works best. When the location for the Balance Due varies (what we call a floating total), draw a long box down the page and choose ‘Closest to the bottom of the box.’
    When skipping a tab, you do not have to position the colored box in a blank space on the template image. You may skip ‘Invoice Number’, for example, if your invoice numbering schedule is so unique it does not match any of the options in Capture.
    Drag the mouse from left to right to easily highlight multiple words.
    To remove highlighted text, simply highlight the text again and mouse click.
    Use the Ctrl + mouse wheel, like you do in TimberScan, to zoom in and out of an image.
    Once a document has been successfully recognized and moved to TimberScan data entry, select the F8 key to display a log of Capture notes.
    Capture maintains a log file similar to the TimberScan user log. These files are located in the Timberscan\LogFiles directory on the server. The file name begins with the TimberScan user ID, followed by OCR and the date. The extension is log.
There are three reasons a document will move to the ‘Unrecognized Capture Documents’ queue:
1) The template is not setup or recognized.
     2) A coding profile is not found
     3) The invoice already exists (duplicate invoice)    
The help manual was created with Dr.Explain