Test Template



TIPS:


Few tips below:

 

  1. You will only need to “map” the invoice (tell Inn-Flow what information goes where) one time per vendor. It should learn this next time you extract an invoice from the same vendor, so you will not need to click the gear icon.

 

  1. It can currently only “read” 1 page of an invoice, so I recommend using the first page so it picks up the invoice number, date, etc.

 

  1. If a value does not appear (like the Due Date), leave that row alone as shown below so nothing imports. The same thing goes if an invoice has WAY too much descriptions, like a Sysco or US Food invoice. You may not want to import all of the descriptions, so you can leave these blank and only import the top portion of the invoice image.

 

 

  1. Since it does import multiple descriptions, there’s now the option to Assign the same COA to multiple rows. While you’re using this, try completing this step. It will also start to read and code the item based on the description. This part is really cool! You just have to start using it for it to start picking this up. ?

 

A screenshot of a computer Description automatically generated



Understanding the Limitations of Automated Invoice Extraction

We are dedicated to employing advanced technologies to enhance your operations and increase efficiency. As a part of our offering, we utilize optical character recognition (OCR) and natural language processing (NLP) to save your team from time-consuming manual data entry.

While this technology is advanced, we want you to be aware that automated invoice extraction isn't flawless. There are several factors that can influence the accuracy of the extraction, and understanding these can help set realistic expectations and minimize potential frustrations.

Here are some factors that might impact the performance:

  1. Poor Image Quality: The quality of the invoice image is a crucial factor. Blurry images, skewed angles, or images with poor contrast may pose challenges for the extraction system. High-quality, clear, and straight-on images will yield the most accurate results.
  2. Unclear Tables or Layouts: Automated systems rely on recognizing common patterns or layouts in invoices. If an invoice's table is not clearly defined or if the layout is unusual, the system may struggle to correctly identify and extract all fields.
  3. Complex or Multi-page Invoices: Invoices that span multiple pages or that have complex structures can sometimes confuse the extraction system, leading to missed fields.
  4. Handwritten or Cursive Fonts: Automated systems excel at recognizing printed text, but they may struggle with handwriting or unusual fonts.
  5. Unusual or Rare Terms: Our system is trained to recognize common invoice terminology. However, if your invoices contain unique or industry-specific terms, it might fail to correctly extract these.
  6. Foreign Languages or Special Characters: While our system supports multiple languages, invoices in languages other than English or those that use special characters may sometimes pose challenges.
  7. Low Resolution or Faded Print: Low-resolution images or faded prints can make it difficult for the system to distinguish between different characters and numbers, leading to inaccuracies in the extracted data.

While these challenges may occur, they tend to be exceptions rather than the rule. We continually monitor our systems to offer you the most accurate results possible. Your understanding and patience with the occasional error is greatly appreciated.