# PDF to Searchable PDF: What Carries Over and What Changes

- Author: Schema Law (https://www.schemalaw.com/)
- Published: 2025-04-08T10:07:21Z
- Updated: 2026-10-11T19:22:20Z
- Category: Searchable PDFs
- Canonical: https://www.schemalaw.com/blog/pdf-to-searchable-pdf

> Converting a PDF to a searchable PDF adds hidden text. See what can happen to bookmarks, links, comments, form fields and metadata, with a comparison sheet.

Converting a PDF to a searchable PDF is meant to change one thing: recognized text is added behind each page. Whether bookmarks, links, comments, form fields, attachments and document properties come through depends on how the tool builds the new file, so compare the two files before you rely on the converted copy.

A PDF is more than its pages. A file that has been worked on in a law office may carry a bookmark panel, clickable cross-references, sticky notes, fillable fields and a set of properties recording where it came from. Conversion tools differ in how much of that they carry across, and none of it shows when you glance at page one.

Search the converted file for the party names. On an appeal, search the appellant and the appellee in the caption, plus an affidavit line you remember from the source. Open the hit and compare it with the page image. Click the bookmark for that pleading and see whether it opens the same page.

## What does converting a PDF to a searchable PDF add to the file?

Converting a PDF to a searchable PDF adds a text layer: characters that sit with each page picture and that your reader's search can match. In the plainest build nothing else is meant to change. The pages look the same, and the file now answers a search.

### The intended change

OCR reads the picture of each page and writes what it read into the file as hidden text. Your PDF reader searches that text and shows you the picture. The guide to making a scanned PDF searchable explains [what that text layer is](https://www.schemalaw.com/blog/how-to-make-a-scanned-pdf-searchable#what-makes-a-scanned-pdf-searchable). The Schema Law homepage describes its own Searchable PDF export in those terms, as "Your original file with a hidden text layer, so search finds the page", and the [formats section](https://www.schemalaw.com/#formats) lists it beside the seven other exports.

### The changes nobody asked for

Adding text means writing a new file, and a tool has choices about how. Some tools insert the text and leave the rest of the file as it was. Others take a picture of every page, recognize that, and assemble a new PDF from the pictures. The second approach can drop or merge everything that was not part of the page picture.

The documentation of OCRmyPDF, an open-source OCR tool, draws this line itself. Its [introduction](https://ocrmypdf.readthedocs.io/en/latest/introduction.html) says that rasterizing a PDF and running the image through an OCR engine "actually generates a new PDF, potentially resulting in the loss of various details (such as the document's metadata)", and contrasts that with producing "a minimally altered PDF". You cannot tell from the output's first page which kind of tool you used.

### What sits outside the page picture

| Part of the file | Where you see it | Part of the page picture? |
| --- | --- | --- |
| Scanned image of the page | The page itself | Yes |
| Stamps and signatures inked on the paper | The page itself | Yes |
| Bookmarks | The bookmarks panel | No |
| Links | Clickable areas on the page | No |
| Comments, highlights, electronic stamps | On the page and in the comments panel | No |
| Form fields | Fillable boxes on the page | No |
| Attached files | The attachments panel | No |
| Title, author, producer, dates | Document properties | No |

Everything in a "No" row is at the mercy of the conversion. The page picture is its own subject, and the post on [whether OCR changes the original PDF](https://www.schemalaw.com/blog/does-ocr-change-the-original-pdf) covers resolution, compression and rotation.

## Do bookmarks, links and page labels survive conversion to a searchable PDF?

Bookmarks, links and page labels survive conversion only if the tool copies them into the new file. They are stored apart from the page contents, so a tool that rebuilds the document from page pictures has to carry them over deliberately. Open the navigation panel of the converted file and look.

### Bookmarks

A bookmark panel on a long PDF represents somebody's work: an exhibit list, a record on appeal, a closing binder with a tab for each document. Bookmarks can also be generated when the PDF is made. Microsoft's instructions for [saving Office files as PDF](https://support.microsoft.com/en-us/office/collab-files/save-or-convert-to-pdf-or-xps-in-office-desktop-apps) describe a "Create bookmarks using" option that builds them from headings or Word bookmarks.

Test bookmarks in three steps. Count the entries in the original and in the converted file. Expand any nested levels and confirm the nesting is the same. Click the first, the last and one in the middle, and confirm each lands on the same page in both files.

If the bookmarks are gone, you have two choices. Convert with a tool that keeps them, or keep the bookmarked original as the navigation copy and use the searchable copy for search. Rebuilding a long bookmark panel by hand is the expensive way out.

### Links

Links come in two kinds: links to web addresses and links that jump to another page in the same file, such as a table of contents entry. Tools may treat them differently from other items on the page. OCRmyPDF's [advanced documentation](https://ocrmypdf.readthedocs.io/en/latest/advanced.html) says of its mode that rasterizes every page: "Hyperlinks (link annotations) are kept, so links to web pages and to other places in the document still work in the rasterized output". That is one tool's statement about one mode. For yours, click a table of contents entry and a web link in the converted file.

### Page labels

Page labels let a PDF's page counter show "iv" or "A-12" in place of a plain sequence number. If the original used them and the conversion dropped them, the counter reverts to 1, 2, 3 and every "go to page" habit your team has for that file breaks. The post on mismatched page numbers explains [what page labels are](https://www.schemalaw.com/blog/pdf-page-numbers-dont-match#what-are-page-labels-in-a-pdf). To test, go to the same labelled page in both files and compare what the counter shows.

## Comments, highlights and stamps on a PDF that goes through OCR

Comments, highlights and electronic stamps are annotations that sit on top of the page. A conversion can keep them as separate objects, drop them, or flatten them into the page picture so they can no longer be removed. Decide what you want to happen to them before you convert.

### Three outcomes

1. **Kept.** The comments panel of the converted file lists the same items. You can still open, edit and delete each one.
2. **Dropped.** The page looks clean and the comments panel is empty. Your notes are in the original only.
3. **Flattened.** The highlight or note icon is visible on the page, and the comments panel is empty. The mark has become part of the picture.

The third outcome is the one to watch. OCRmyPDF's documentation states it outright for its rasterizing mode: after noting that hyperlinks are kept, it says "Other annotations are flattened into the image." A flattened highlight cannot be turned off. A flattened note icon shows that a note existed. Text that a comment box covered on screen may be covered in the picture.

### Why this matters in a legal file

Annotations on a working copy can record a lawyer's thinking. If a converted copy is going to a client, an expert, the other side or a court, a mark that was flattened into the page travels with it. This post does not address privilege or work product. It only points out that the safe input for a conversion you will share is a copy with no annotations on it.

### Ink stamps are different

A received stamp, a Bates number or an exhibit sticker that was on the paper when it was scanned is part of the page picture and stays where it is. What happens to the print underneath it during recognition is a separate question about text quality, not about what the file keeps.

### What to do

- Convert the clean copy, as received, before anyone marks it up.
- If only an annotated copy exists, save a copy with the annotations removed, convert that, and keep the annotated version for internal use.
- After converting, open the comments panel of the output and scroll two or three marked pages.
- The post on [preparing a PDF for OCR](https://www.schemalaw.com/blog/prepare-a-pdf-for-ocr) covers this and other housekeeping.

## Form fields and attached files after a PDF is converted

Form fields and attached files are interactive content, and a conversion may keep them working, freeze them, or leave them out. A fillable form can come back as a flat picture of its filled-in state. An attachment can be missing from the new file altogether. Check both panels in the output.

### Form fields

A fillable PDF stores each field as an object with a name and a value. The value you see in the box is drawn on top of the page. A tool that takes a picture of each page captures the drawn value and loses the field. OCRmyPDF describes its rasterizing mode as "flattening form fields or interactive objects into their visual representation".

Whether that is a problem depends on what the form is for:

| Situation | Flattened fields are |
| --- | --- |
| A completed form you are filing away as a record | Acceptable, and sometimes wanted, since the entries can no longer be edited |
| A form someone still has to fill in or sign | A failure; the boxes no longer accept typing |
| A form whose data another system will read from the fields | A failure; the values are now only pixels and hidden text |
| A form with calculated totals | A snapshot; the total will not update |

Test search on a fillable form before you convert it. If its labels and instructions already search, the form holds typed text and does not need OCR. If they do not, find the reason before converting, as set out in the post on [how to make a PDF searchable](https://www.schemalaw.com/blog/how-to-make-a-pdf-searchable).

### Attached files

A PDF can carry other files inside it: a spreadsheet behind an expert report, the email a letter was attached to. Open the attachments panel of the original before conversion and write down what is there. Open the same panel in the output. If the list is shorter, save the attachments out of the original and file them next to the searchable copy.

Recognition does not read attachments. An attached scan is as unsearchable after the parent file is converted as it was before, and needs converting as its own document.

### Signatures

A certificate-based digital signature is a special kind of field, and changing a signed file has consequences for it. The post on [making a signed PDF searchable](https://www.schemalaw.com/blog/make-a-signed-pdf-searchable) deals with that case, and it is the one to read before you convert anything signed that way.

## Why does the converted searchable PDF report a different creator and date?

The converted file reports a different creator and date because a different program wrote it at a different time. Document properties record the software that produced a PDF and when it was made and modified. A conversion writes a new file, and the new file describes its own making.

### What the properties hold

Adobe's [overview of document properties and metadata](https://helpx.adobe.com/acrobat/using/pdf-properties-metadata.html) lists basic information such as title, author, subject and keywords, and "standard metadata fields" such as creation date and modification date. Open the properties of both files side by side and you may find:

- **Producer or application.** The original names a scanner, a copier or an office program. The converted file names the OCR tool.
- **Modified date.** The converted file shows the day you ran the conversion.
- **Created date.** Carried over by some tools, reset by others.
- **Title, author, subject, keywords.** Carried over, blanked, or replaced.

The dates your operating system shows for the file are separate from all of these and will be today's, since the file was saved today.

### What that means for your records

None of this is a fault. It does mean the converted copy cannot stand in for the original when the question is where a document came from or when it was made. Properties that matter to the matter belong to the file as received. Keep that file untouched, and treat the searchable copy as a working copy whose properties describe your conversion.

Two practical points follow. If your office sorts or filters documents by PDF properties, the converted copies will sort by conversion date unless the tool carried the old values across. And if you send the searchable copy outside the office, its properties name the tool you used and the date you used it. Look at them before sending, as you would with any outgoing file.

## A before-and-after comparison sheet for a converted PDF

A comparison sheet lists each part of the file, what the original had, what the converted copy has, and whether the difference matters. Fill it in once for each tool you use, on a file that exercises every row, and again for any single file where navigation or annotations matter.

Open the original and the converted copy in two windows and fill in each row.

| Item | Original | Converted | Match? |
| --- | --- | --- | --- |
| Page count | | | |
| Search for a visible word, late page | No hit expected | Hit expected | |
| Number of bookmarks | | | |
| Middle bookmark lands on page | | | |
| Internal link (contents entry) works | | | |
| Web link works | | | |
| Page label shown on a labelled page | | | |
| Number of comments in the panel | | | |
| Any annotation visible on the page but absent from the panel | | | |
| A form field accepts typing | | | |
| Number of attachments | | | |
| Signature panel entries | | | |
| Producer or application in properties | | | |
| Created and modified dates | | | |
| Title and author | | | |
| File size | | | |

### Reading the sheet

A "no" in a row is a decision, not automatically a defect. Flattened fields on a completed form may be fine. Missing bookmarks on a 400-page record are not. For each mismatch, write one of three outcomes: accepted, fixed by another route, or original kept as the copy for that purpose.

### Building a test file

To learn how a tool behaves before a real file depends on it, make a five-page test PDF with no client content: one bookmark, one internal link, one web link, one sticky note, one highlight, one fillable field and one small attachment. Convert it and fill in the sheet. That one file tells you what the tool carries over, and you can repeat the test whenever the tool is updated.

Keep the completed sheet with your notes on the tool. When a colleague asks whether it is safe to convert the bookmarked appendix, the answer is already written down. For the route from each kind of input to a finished file, the post on [creating a searchable PDF](https://www.schemalaw.com/blog/create-searchable-pdf) is the place to start.

## Try it on a real document

Upload a scanned PDF and export it as Word, a searchable PDF or plain text on the free plan. Your documents are never used to train an AI model. [Start free](https://www.schemalaw.com/api/auth/google)