Searchable scanned PDFs: what browser tools can and cannot do

A flat scan PDF is a stack of images. You can scroll and print, but you cannot select sentences. Search in the viewer fails. Accessibility tools read nothing. “Searchable PDF” means a hidden text layer aligns with the pictures. Building that layer is OCR plus PDF assembly.

QuickConvert does not yet inject invisible text layers into PDFs in one click. You can still assemble a workflow with Image OCR, PDF to Images, and Images to PDF that stays in the browser.

Export pages as images

Open a stubborn scan in your viewer or use PDF to Images to dump pages as PNG. Clear scans OCR better than blurry embedded JPEGs inside old PDFs.

Run OCR per page or per cropped column. Paste text into a structured document if you need chapters. For quick find-in-notes, saving plain .txt per page may be enough.

Some teams paste OCR text under the image in Word, then export with Office to PDF. That yields selectable text without a perfect invisible layer, but search may work. Match fonts approximately so reflow does not surprise readers.

When exhibits must stay image-only, keep OCR text in a sidecar spreadsheet with page numbers.

Merge after processing

Chapter scans merged via PDF Merge remain image-only unless you add text elsewhere. Plan OCR before merge so filenames map to sections.

Split oversized batch scans using the separate-scanned-documents guide before OCR so each logical document gets its own text file.

Scan at 300 dpi, grayscale or colour as needed, with deskew on. Saving as TIFF on the scanner then converting for web is fine. Avoid dual-page spreads if OCR should run per page.

Phone photos can work with good light and Image Crop. Expect more editing than a flatbed.

Screen readers need text tags, headings, and reading order. OCR text dumped without structure is better than silence but not equal to tagged PDF. For public publishing, use desktop tagging tools after OCR draft.

Read the companion post on PDF accessibility basics for headings and alt text thinking even when you only have scans.

Tax and medical scans should not go to cloud OCR if you can avoid it. Local OCR on QuickConvert keeps bytes on device. Store outputs encrypted on disk if required by policy.

Huge historical archives may justify commercial OCR with auditing. For personal batches under a hundred pages, browser prep plus careful OCR often suffices.

Compress scans only after OCR if size matters; aggressive JPEG hurts recognition. PDF Compress helps emailed archives after text is captured elsewhere.

Does QuickConvert create searchable PDF automatically? Not as a hidden layer product. Use OCR text output and rebuild as needed.

Can I search inside the image PDF? No until a text layer exists. Some viewers fake search on OCR sidecars they manage separately.

Will OCR preserve tables? Rarely. Rebuild tables manually or in Excel.

Should I OCR before or after compress? OCR before heavy compress. Compress for sharing after text is safe elsewhere.

Can I OCR a password PDF? Unlock first. Tools refuse encrypted inputs.

What about multilingual scans? OCR per language section after crop.

How do I cite page numbers? Keep a manifest mapping page image to OCR file names.

Is plain text enough for lawyers? Sometimes for discovery search, but exhibits stay image PDFs. Ask the requesting party.

Some teams keep a git repo of .txt OCR outputs keyed by SHA-256 of the scan PDF. Search ripgrep across text files when lawyers ask for a clause. The image PDF stays evidence; the text repo is the working index. QuickConvert’s OCR export feeds that pattern without uploading either file.

Document which pages were re-scanned after corrections so OCR files get regenerated in lockstep.

When discovery asks for “all text”, export your sidecar index with page pointers rather than promising magic inside the image PDF.

Historical archives

Family historians scanning letters should OCR for names and places, then keep albums in Images to PDF for relatives who will never open a .txt file. Both outputs from the same browser session respect privacy for faded ink on personal stationery.

Medical records

Clinic faxes scanned to PDF rarely arrive searchable. Nurses can OCR medication lists into the EHR while the image scan satisfies retention. Validate drug names manually; OCR confuses mg and mg once a shift.

Academic reading

Students photographing library pages should OCR nightly notes before the book goes back on shelf. Paste quotes with page labels into study guides. Citation still needs the physical book; OCR is a study aid.

Limits of find-in-PDF

Desktop viewers that claim search on scans sometimes run hidden cloud OCR. QuickConvert’s path keeps that step explicit on your hardware. Read PDF to Images docs on the tool page when embedded JPEGs look worse than fresh exports.

Budget time: a 200-page scan is two hundred thoughtful OCR passes or a farm job elsewhere. Browser OCR suits chapters, not warehouses.

Libraries publishing public-domain scans should pair downloadable image PDFs with plain-text transcriptions in the catalog record, not assume Google will OCR for you.

Journalists photographing source documents should OCR quotes before deadline while retaining the photo PDF for the legal team’s verification pass.

Court clerks digitising dockets should note in the CMS which volumes were OCR’d locally versus vendor batches so appeals counsel knows where to spot-check.