A scanned page may look like text while still behaving like an image. In Acrobat 2025 Pro, optical character recognition can add searchable text, but the result still needs review.
Start with a usable scan
Check that the page is complete, reasonably straight and readable. A cut-off margin or blurred reference number cannot be reliably reconstructed simply by running recognition. Keep the original scan so the recognised result can be compared with it.
Adobe’s procedure uses the Scan & OCR tools, with a choice of page range and recognition language. Acrobat then creates a searchable text layer. Choose those settings deliberately, particularly for documents containing more than one language.
Check the recognition result
- Run recognition on a representative page or the intended page range.
- Search for a known word from the document.
- Inspect names, dates, reference numbers and decimal values.
- Check a difficult area such as a stamp, table or low-contrast paragraph.
- Save a clearly named output while retaining the source scan.
Use the text appropriately
Searchability makes a document easier to navigate; it does not prove that every recognised character is correct. If the text will be used in another process, verify the fields that matter to that process before relying on them.
For example, an invoice reference containing the letter O and the number 0 deserves a direct comparison with the image. A successful search for the supplier’s name does not validate the reference or amount.
Keep the review proportionate to the use of the document and record any remaining recognition problems for the next person working with it.
Publisher reference: Adobe: recognise text in scanned PDFs.
Check the platform and product details on the Acrobat 2025 Pro listing before preparing your PDF workflow.