Copying information from source documents is an extremely difficult task. As you said, it may lead to erroneous data imports that affect data analysis and data quality.
Let’s me share with you, I have experience working with CRF too, which is copying data from source documents. After that, I will import them into the computer system. The system will use an intelligent character recognition (ICR) software to read the data on it, then trying to import to the database.
It also depends on the CRF design. If it is designed with mostly checkbox fields, the software will read them almost 100% correctly. However, if fields are handwriting, there might be a higher error in reading.
Thanks for sharing,