Our Optical Character Recognition Best Practices

When you outsource your data management challenges to DataEntryOutsourced, DEO exercises constant vigilance in applying best practices to optical character recognition analysis and data conversion. We are proud of our best practices because they enable us to help you achieve greater success — and your success defines our success.

Our Best Practices

  • Understanding the Source Material

    We start by assessing the printed source material for unique characteristics that will make the conversion process either simpler or more difficult. For example, OCR applications depend on lexical data and existing modern-language dictionaries — with special terminology and historical language, the required lexical data might not exist. Project goals play a key role in defining the most efficient way to automate the process of filtering pages for special OCR treatment.

Creating a Process Workflow

Without a viable plan, the OCR data conversion process will be chaotic. Failure to use a highly-disciplined approach to workflow will ultimately disrupt a successful conversion and lead to other pitfalls. DataEntryOutsourced streamlines everything before starting — DEO steps include the following:

  • Document registration
  • Verification of documents
  • Attribution and analytical description of the document book
  • Archiving and multi-acceptance of documents with attachments
  • Quality assurance that includes manual checking

While perfection can be a worthy goal, the expected accuracy rate for DEO data conversion processes using optical character recognition is approximately 95 percent. This projected window of potential inaccuracies places a special premium on manual checking at the end — with our commitment to quality, we will tweak the data to get the best possible results.

No One-Size-Fits-All Approach

Avoiding a "one-size-fits-all" mentality is mandatory for data conversion success. Each part of a data conversion project must treated as unique — what works in one case will often not work in another. Keep in mind that in some OCR implementations, making changes may demonstrate improvement in one area and detriment in another. The most likely case is settings for one document type are not working well for another." (Excerpt from "The New Best Practices for Document Imaging, OCR and Scanning)

Manual revision of each document is an effective but unworkable and cost-prohibitive way to ensure accuracy of data conversions. DEO uses a simple but effective alternative — using a random batch of files for manually checking the software log of an OCR engine. The software log is where the software keeps score, and it is prudent to make use of this information when calculating success rates.

Avoiding Pitfalls

DataEntryOutsourced is committed to ensuring that your OCR data conversion projects avoid common pitfalls. Here are two key points to remember:

  • Unrealistic expectations — A more realistic expectation is in the range of 85 to 95 percent accuracy. This is actually higher than the accuracy when an individual retypes content piece-by-piece.
  • Failure to define needs — The biggest factor for the success or failure of an OCR project is the quality of inputs you provide including the original documents. It is vital to give accurate inputs and define the project goals correctly to see the desired results achieved.

Ensuring Effective ROI

Reducing expenses and increasing cash flow can both be achieved if you carefully plan and execute an OCR data conversion project. Whatever you choose to invest in document imaging and scanning can produce an attractive return on investment — often in unexpected ways. For example, if you are able to process invoices more quickly, reduce staff and eliminate manual data entry, you are likely to see an immediate bump in your ROI. But that should be just the beginning — the ability to quickly and easily access, manage and edit data is "priceless" for most successful businesses.

Should you outsource OCR data conversion to DataEntryOutsourced?

At a minimum, you should review the costs and benefits. Without the need to hire a single employee, hiring DEO allows you to have an entire team of data management specialists working for you immediately. Does your company have the capacity to use the OCR technology, and if your answer is no, is it possible to outsource, or do possibilities exist in developing capacity in-house in the immediate future, are some of the key strategy questions you and your team may want to consider.

Ask for a Free Trial!