How to make a PDF searchable in Adobe Acrobat — or free, without it
In paid Acrobat, open the scan, choose Scan & OCR, then Recognize text, and save. Without Acrobat, OCR PDF does it in your browser. Either way, run OCR before adding page numbers or a watermark: one stamped word blocks the page.
Acrobat makes a scan searchable with its Recognize Text command, part of the paid desktop app. This covers that route as Adobe describes it, the free route in a browser, and the one mistake that blocks OCR in both: putting real text on the page first.
How do I get Adobe to recognize text?
Open the scanned PDF in desktop Acrobat, choose the Scan & OCR tool, then Recognize text, and save the result. Adobe's searchable-PDF guide picks the output style Searchable Image before you confirm. These are Adobe's steps; we have not tested them inside Acrobat.
Two other Adobe routes exist. Choosing Edit on a scan makes Acrobat run OCR on its own and turn the page into an editable copy, according to Adobe's help page. Adobe's free online OCR tool says your file is handled by Adobe servers, and it asks you to sign in before you can download the result.
How can I make a PDF searchable without Acrobat?
Open OCR PDF in any browser, choose the scan, and click Make PDF searchable. It is free, needs no account, and the recognition runs in the tab, so the file is not uploaded. The copy downloads with _searchable added to its name.
Open OCR PDF
Go to OCR PDF in any current browser. The first run downloads the roughly 7 MB recognition engine from this site, and your browser keeps it cached after that.
Choose the scanned PDF
Drop the file in or click to pick it. A password-protected PDF has to be unlocked first.
Click Make PDF searchable
Each page is rendered and read in turn, with a page counter as it goes. You can switch to another tab while it works.
Check the report
The searchable copy downloads by itself. The tool shows how many pages it read, how many words it added, the engine's confidence score, and any pages it left alone.
Why isn't my PDF recognizing text?
One common reason is that the page already has real text on it, even a single word. Acrobat refuses such a page with "This page contains renderable text", and Adobe's support page says Acrobat cannot run OCR on renderable text. OCR PDF has the same rule: a page with any selectable text is skipped.
We tested how little it takes. We made a three-page scan with no text layer, ran it through Add Page Numbers, and then tried OCR. A single number in the footer of each page was enough: every page counted as already searchable, and the tool refused the whole file with "Every page in this PDF already has selectable text". A COPY watermark from Add Watermark did the same. When only the middle page carried a number, the first and last pages were read and the middle one was skipped.
The skip is on purpose. Reading a page that already has text would lay a second, slightly wrong copy of every word under the real one. The cost is that a scan with a page number, a Bates stamp or a watermark added afterwards by software looks searchable to the tool when almost none of it is.
How do I OCR a PDF that already has a stamp on it?
Run OCR on a copy made before the stamp, then add the page numbers or watermark again. Adobe gives the same first advice. In our test, OCR followed by Add Page Numbers gave pages where the recognised text and the new numbers were both searchable.
If there is no clean copy, Adobe suggests saving the PDF as TIFF images and recognising those. The closest route here is Flatten PDF with Everything to images, then OCR. It worked in our test, and OCR read the page numbers back too. It costs something: flattening drops the title and other document details and stores each page as a new JPEG picture, so the scan is re-compressed once. How to flatten a PDF explains that trade.
Why isn't my Adobe search working in a PDF?
Search reads text, and a scan has none until OCR adds it, which is why the cursor finds nothing. After OCR, a search can still miss a word on a page that was skipped, a word the engine scored below 30% confidence, or anything outside English.
OCR PDF drops those low-confidence words instead of hiding them in the file, and it counts them in its report, so you know where the gaps are. The recognition model is English only.
What does a searchable copy change in the file?
Only the text layer. We compared the scan images before and after OCR by checksum, and they were identical: nothing was re-encoded. The recognised words are drawn over the picture with their fill set to fully transparent. Our three-page test grew from 95,016 to 102,326 bytes for 61 recognised words.
That also means the result is searchable, not editable. Acrobat's Edit mode rebuilds a scan into text you can retype; OCR PDF does not. For more, see how to make a scanned PDF searchable and what OCR can and cannot recover.
