OCR recognises characters in page images. It can make a scan searchable, but its accuracy depends on the original image and language.
Improve the scan first
Use a straight, well-lit scan with sharp text and a plain background. Remove excessive borders where possible. Very small type, shadows and motion blur can produce plausible but incorrect words. Handwriting recognition is especially unreliable.
Choose the correct language
Use OCR PDF with the relevant supported language. This service supports English and Hindi processing where available in the workspace. It temporarily uploads the document only after you agree to server processing. Avoid uploading material you are not permitted to share with the service.
Review important values
Compare names, amounts, reference numbers and dates against the original. OCR can confuse 0 with O, 1 with l, or merge columns into an incorrect reading order. Searchability does not mean the extracted text is authoritative.
Keep the page image
Use the searchable PDF as a convenience copy and retain the original scan. If you export text or Markdown, inspect headings, lists and tables separately. Complex page layouts often need manual correction before reuse.