Do I need OCR on scanned documents?

Scanning a paper document produces an image, a picture of the page. By itself, that image is not searchable or selectable as text; a computer sees pixels, not words. Optical Character Recognition (OCR) is the process that analyzes the image and creates a machine-readable text layer behind it. The short answer is that OCR is not strictly required to have a legitimate digital record, but in most modern recordkeeping programs it is highly recommended.

When OCR Is Worth It

OCR delivers the most value when documents need to be:

• Found by full-text search. Without a text layer, users can only locate records by file name or metadata. OCR lets people search the actual words inside thousands of pages.
• Accessible. A text layer allows screen readers to voice the content, supporting people with visual impairments and helping meet accessibility expectations.
• Reused or analyzed. Copy-paste, redaction tools, data extraction, and e-discovery review all rely on selectable text.
• Responsive to requests. For FOIA, audits, or litigation, searchable text dramatically speeds review and production.

When You Might Skip It

OCR is less essential when records are rarely retrieved, are already well described by metadata, or contain content OCR handles poorly, such as handwriting, complex forms, or degraded originals. In those cases the image may be the authoritative copy and OCR adds little. Even so, capturing good descriptive metadata at scan time remains important.

Quality Considerations

OCR accuracy depends heavily on scan quality. Following established imaging targets, such as appropriate resolution, color, and quality control, produces cleaner images and far better OCR results. Plan to validate output, especially for high-value or permanent records, because OCR is rarely perfect.

A common, practical approach is to store a searchable PDF: the page image preserves the visual record while an embedded text layer supports search and access. Importantly, OCR does not replace the original image, and for records of long-term or permanent value you should preserve the image as the trustworthy representation rather than relying on the OCR text alone.

For broader guidance on planning a digitization effort, see the digitization and imaging topic hub.