Modern bank statements are usually text-based PDFs — the text is selectable and copyable. But some bank statements come as scanned images: photographs of paper statements, statements from older banking systems that print to image PDF, or screenshots from mobile banking apps.
Image PDFs require an extra OCR step before any converter can extract the data. The quality of the OCR determines the quality of the final output. Here's how to handle these cleanly.
Short version: Identify whether your PDF is text-based or image-based first (try selecting text — if you can highlight it, it's text). For image PDFs, run them through a dedicated OCR step (Adobe Acrobat, ABBYY FineReader, or a bank-statement-aware converter with built-in OCR) before extracting transactions. Quality of source image matters enormously.
Identifying scanned vs text PDFs
The fastest test: open the PDF and try to select text. If you can highlight individual words, it's a text PDF. If selecting just highlights blocks or the whole page as an image, it's a scanned PDF.
Other signs you have a scanned PDF:
- Visible scan artifacts — slight skew, dust marks, uneven brightness
- Edges of the page show physical paper edges
- Large file size relative to page count (image PDFs are bigger than text PDFs)
- Text quality varies across the page
Why scanned PDFs are harder to convert
Text PDFs contain the actual text characters in the file. A converter just reads them. Scanned PDFs contain only images — the converter must first OCR the image to extract text, then parse the extracted text into transactions.
OCR introduces errors. A "5" might become a "6," an "8" might become a "B." Numbers misread as letters break amount parsing. Letters misread as numbers break date parsing. The error rate scales with scan quality.
The workflow for handling scanned PDFs
Step 1: Improve the source if possible
If your client emailed you a low-quality scan and you can request a better one, do so. Sources to ask for, in order of quality:
- The original text PDF from the bank's online banking portal (most banks let you download statements as text PDFs)
- A direct CSV export from the bank's online portal
- A high-resolution scan (300+ DPI, scanned, not photographed)
- A photograph taken in good light, flat against the page
If you can get #1 or #2, you've sidestepped the OCR problem entirely.
Step 2: Pre-process the image if it's bad
For photographs or low-quality scans:
- Use a tool like Adobe Acrobat's "Optimize" feature to enhance contrast
- Straighten skewed pages
- Crop out anything outside the statement boundaries
This is overkill for good-quality scans but essential for marginal ones.
Step 3: OCR the PDF
Three main options:
- Adobe Acrobat Pro: Built-in OCR is decent for clean scans. The "Make PDF Searchable" feature creates a hybrid PDF with both the image and extracted text.
- ABBYY FineReader: The accuracy leader for difficult documents. Expensive but worth it for high-volume scan work.
- OCR built into a converter: Some bank statement converters (MoneyThumb, certain modes of DocuClipper) have integrated OCR. Quality varies.
Step 4: Verify the OCR output
Open the OCR'd PDF and spot-check 10-20 transactions. Are the amounts right? Are the dates right? If you see consistent errors (e.g., "0" misread as "O" in amounts), the OCR isn't reliable for this document. Try a different OCR engine or a higher-quality source.
Step 5: Run the OCR'd PDF through your converter
Now that the PDF has actual text, run it through a normal bank statement converter. The output should be reasonable as long as the OCR was good.
Step 6: Reconcile and review
Always reconcile a scan-derived file against the statement totals before importing to QBO. OCR errors compound, and you want to catch them at conversion, not after import. See our QBO reconciliation guide.
What MoneyThumb and similar legacy tools do well
If you regularly handle scanned PDFs, it's worth knowing that MoneyThumb's OCR is among the best in this category — particularly on older bank statements and small credit union formats. The UI is dated, but the OCR layer is solid.
If your firm handles a lot of older clients or clients at small banks that still mail paper statements, MoneyThumb might be worth the cost specifically for the OCR capability.
When to refuse the engagement
If a client hands you 24 months of low-quality phone-photo statements with no online banking access available, the reconstruction work is significant and should be priced accordingly. Some engagements aren't worth taking at any price — pages where the OCR can't reliably extract the data require manual keying, which defeats the value of the converter.
Politely decline or quote at hourly rates for the manual portion.
For text-based PDFs (the easier case): YourStatementConverter handles modern text PDFs with built-in reconciliation. 25 pages free, no credit card. For scanned PDFs, OCR them first.