What is OCR?

In practice

Modern OCR handles clean scans of ordinary documents very well and still struggles with the same things it always has: handwriting, poor photographs, stamps over text, dense multi-column layouts and complex tables. Knowing whether your documents are born-digital PDFs or scans changes the cost of a project substantially, and it is worth checking before anyone estimates.

Where it fits in a build

OCR belongs at the front of any document workflow that receives paper or images. Where files are born digital, skipping OCR and reading the embedded text directly is both more accurate and cheaper.

Common mistakes

• Running OCR over PDFs that already contain text, which degrades quality for no reason.
• Not keeping the position of each extracted value, so nobody can point at where in the document it came from.
• Assuming a table survived extraction because the numbers did.

Related terms

• Document intelligence
• Business process automation (BPA)
• Audit trail

Working on something that involves this?

ASTACKRA designs and builds AI systems, automation and custom software for businesses that need technology shaped around their own workflow. If this term turned up in a proposal and you want a straight answer about whether it applies to your situation, ask us — no obligation, and we will tell you if the answer is no.