To get a PDF image to Excel, first confirm that it is an image: if you cannot select a word on the page, it is a picture of text. Excel's PDF import cannot read a picture, so the file has to pass through OCR, which adds a text layer, before the numbers can reach cells.
How do you tell an image PDF from a text PDF?
An image PDF holds each page as a picture, while a text PDF holds the actual characters. You can tell them apart in about ten seconds with three checks: try to select a word, search for a word you can see, and look at the document properties for a scanner as the producer.
Run the checks in this order. Stop as soon as one gives a clear answer.
- Try to select a word. Open the PDF in any reader and drag across a line of text, or double-click a word. On a text PDF the word highlights, and you can copy it. On an image PDF nothing highlights, or the whole page highlights as one block.
- Search for a word you can see. Press Ctrl+F (Cmd+F on a Mac) and type a word that is plainly on the page, such as "Balance". A text PDF finds it. An image PDF reports zero matches even though you are looking right at the word.
- Check the document properties. In most readers, open File, then Properties or Document Properties. The "Producer" or "Creator" field often names a scanner, a copier or a scanning app when the file started as paper. A name of a bank's statement system or a word processor suggests real text.
A quick result guide:
| Select a word | Search finds it | Producer field | Likely type |
|---|---|---|---|
| Yes | Yes | Document software | Text PDF |
| No | No | Scanner or scan app | Image PDF |
| Yes | Yes | Scanner or scan app | Image PDF with a text layer added |
| Some pages only | Some pages only | Mixed | Mixed PDF |
The third row is worth a note. Many scanners and copiers run OCR themselves and save the result as a "searchable PDF", where the page picture stays on top and the recognized text sits invisibly underneath. That file passes the select-and-search tests, yet its text came from OCR and can contain the digit mistakes that OCR makes. If the producer is a scanner and the numbers matter, check them as you would an OCR result. The digit errors OCR makes on bank statements lists what to look for.
A fourth sign: the file is heavy and the text looks soft
Zoom in to 400 percent. Text PDFs stay crisp at any zoom, because the characters are drawn from outlines. An image PDF turns blocky or blurry because you are enlarging pixels. A statement of a few pages that weighs several megabytes is also a hint, since page pictures take far more space than text.
Why image PDF files come out blank in Excel
Image PDF files come out blank because Excel looks for text and table structure inside the PDF, and a picture has neither. There are no characters to place in cells, so the import reports that no tables were found or returns an empty result.
Excel's Get Data from PDF feature and most Save As routes work the same way. They read the text objects stored in the PDF, work out where rows and columns are from the positions of those objects, and fill cells. An image PDF has one object per page, a picture, and that is all the importer sees. The page looks full of figures to you, and to the importer it is one large block with nothing to split.
You may see one of several symptoms:
- An empty Navigator. The list of tables in the import window has no entries, or only a page icon with nothing in it.
- A picture in a cell. Some routes drop the whole page into the sheet as an image you can resize but not sum.
- A single column of odd characters. If the file has a poor, partial text layer, you may get fragments instead of clean rows.
- A message saying no tables were found.
Each symptom points to the same cause, and none of them is a problem with Excel. Trying a different Excel feature or a different save format will not help, because the missing ingredient is text, not a setting. The routes that work for ordinary PDFs are in saving a PDF as an Excel file. Use them once your file passes the text test.
Mixed PDF files with image and text pages
A mixed PDF has some pages that hold real text and others that hold pictures. It happens when someone combines a downloaded statement with a scanned page, or when a scan of a missing page is added to a digital file. The conversion looks fine until it reaches the picture pages.
The first sign is usually a short result. A twelve-page file produces ten pages of rows, and two pages simply do not appear. Or the running balance breaks at a certain page and never recovers.
To find the picture pages, use the select test page by page. Open the PDF, jump to the first page and try to highlight a word, then do the same at the last page and at a few in between. If a page will not highlight, note its number. Many readers also show page thumbnails, and picture pages often look slightly different in sharpness or margin than the text pages.
You then have two choices:
- Run OCR on the whole file. Most OCR tools skip pages that already have text, or redo them, depending on the setting. Choose the option that keeps existing text and recognizes only the image pages if you have one.
- Split the file. Pull out the picture pages, run them through OCR on their own, and convert the text pages in the normal way. Then join the two results in Excel.
Whichever you pick, check the join. A statement that crosses from a text page to a picture page needs the balance to carry over without a gap. A mixed file is the case where the running balance earns its keep, because the check passes on one page and fails on the next.
Can you add a text layer to an image PDF?
Yes. OCR can add an invisible text layer behind each page picture, which turns the image PDF into a searchable one. Once the layer exists, the PDF behaves like a text PDF for selecting, searching and copying, and a converter can read the characters.
The steps are the same in every tool:
- Open the image PDF in an OCR-capable tool, such as a PDF editor with a recognize-text feature, a scanner program or an online OCR service.
- Choose the language of the document. A wrong language lowers accuracy on names and descriptions.
- Run recognition on all pages, or on only the picture pages in a mixed file.
- Save the result as a searchable PDF.
- Repeat the three-step test on the output. If a word now highlights and searches, the layer is in place.
- Convert to Excel and check the figures.
The text layer is a guess, not a copy. OCR may have misread a digit, and the new layer will carry that error faithfully into Excel. For a bank statement, run the balance check after the conversion rather than assuming the layer is right. For how the recognition step builds rows and columns from the text it finds, see how OCR turns a PDF into Excel columns.
The step of running the scanned file through a converter is covered in running the scanned PDF through an OCR converter, so this post stops at confirming the layer.
Handwriting is the limit of this approach. OCR built for printed type does not read handwritten pages reliably, as what is realistic with a handwritten PDF explains.
Image PDF files from bank portals and email
Banks send image PDFs for a handful of ordinary reasons, and most of them are not signs of a problem with your account. The statement may have been printed and rescanned, produced by an older document system, or flattened to stop it being edited.
The common sources:
- Mailed paper statements that you or a service scanned.
- Statements saved from a browser print dialog as "Print to PDF" in a mode that rasterizes the page, or screenshots pasted into a document.
- Email attachments from a person, not from the bank. A colleague who photographs or scans a page before forwarding it sends an image.
- Archive documents such as older years' statements from a bank's records, which some banks store as scanned images.
- Flattened files, where a tool merged the layers to shrink or protect the file.
If the file came from online banking, try downloading it again by another route before you accept an image. Look for a different button, such as "Download PDF" rather than "Print", or a date-range export for the same period. Many bank sites also offer CSV downloads of transactions for recent months, which skips conversion altogether. The text PDF is the better file whenever the bank can provide it.
When you can only get the image, treat OCR as a required first step, not an optional one. FastBankConvert is aimed at statement PDFs and checks that the running balance adds up before you download. It is still getting ready to open, and its FAQ notes that scanned paper statements take longer than digital ones. Get started to join the list for when trial conversions open.