We deal with a lot of scanned PDFs — old contracts, intake forms, that kind of thing. I’ve tried a couple of OCR tools but the results have been all over the place, especially when the formatting matters. Some stuff comes out as a wall of text with no structure. Is there a go-to that people actually trust for this? Ideally something that keeps tables and layout intact, not just dumps raw text.
PDF OCR is a bit of a minefield, mainly because “PDF” covers really different things. A PDF that was exported from Word already has text baked in — you don’t actually need OCR for that, just extraction. But a scanned PDF is basically a photo, and that’s where real OCR comes in. Mixed documents (some pages digital, some scanned) are super common and tend to trip up a lot of tools.
For formatted output — like you want the result to actually look like the original — ABBYY FineReader is still the gold standard in my experience. Accuracy is excellent, it handles both PDF types, and it preserves layout better than anything else I’ve tested. Adobe Acrobat Pro is solid too if you’re already paying for it, and the OCR is good enough for most use cases. Tesseract is free and open-source; it works fine on clean scans but don’t expect it to maintain any formatting.
FWIW, if your end goal is pulling structured data out of PDFs — invoice fields, table values, form data — rather than producing a readable document, the approach is a bit different. Tools like Lido are built for that use case specifically. It handles scanned and digital PDFs and just gives you the data, already connected to wherever you need it (Sheets, Excel, etc.). No formatting preservation, but if you’re extracting data you don’t really need that anyway.
So honestly it depends what “works best” means for you. Readable documents → ABBYY or Acrobat. Raw data extraction → something like Lido. Need both → ABBYY for the reading, separate extraction layer for the data. Most setups end up combining tools once the volume gets real.
Just to throw in a real-world data point here — we’re about 100 people and pushing somewhere around 1200 invoices a month through the system. We started with Tesseract because, well, free is free. But the accuracy on anything that wasn’t a clean digital PDF was pretty rough. We were sitting around 60-70% on the messier stuff which just wasn’t workable. Switched over to ABBYY a while back and we’re consistently above 95% now. Night and day difference, especially on scanned docs.
Jumping in here because we literally just finished a three-month pilot doing almost exactly this. Tested four different tools. ABBYY came out on top for us, and honestly the thing that sealed it was how smoothly it syncs with Google Sheets — our AP team refuses to leave that environment and I don’t blame them. That was basically a hard requirement going in.