How accurate is OCR?
OCR works best on clear, upright, high-contrast scans. Blurry photos and handwriting reduce accuracy.
Overview
When Sera reads PDFs and images, optical character recognition (OCR) extracts text so Timeline, search, and Ask Sera can use it. Accuracy depends on image quality, language clarity, and layout complexity.
Why this feature exists
Understanding OCR limits helps you decide when to re-scan versus when to trust extracted text.
Step-by-step
- 1
Prefer clear scans
Flat, well-lit pages with readable fonts produce better results than angled phone photos.
- 2
Spot-check important pages
Open key documents and confirm names, dates, and amounts before relying on them in exports.
Best practices
- Re-upload sharper versions if OCR looks garbled.
- Add a short note on Timeline when a scan is unreliable but the event still matters.
Productivity tips
- Batch-scan multipage PDFs when possible instead of many single phone photos.
Common mistakes
- Assuming every handwritten note will OCR perfectly.
- Ignoring processing failures after a bad scan.
Frequently asked questions
How accurate is OCR?
It varies. Clean typed PDFs are usually strong; photos of crumpled paper, low light, or heavy handwriting are weaker. Always verify critical facts against the original file.
What if OCR fails?
Try a higher-quality scan or a text-based PDF. See troubleshooting for OCR problems.
Screenshots
Illustration placeholder — product screenshot coming soon.
How accurate is OCR? screenshot placeholderHelpful links
Related articles
- OCR problems
Improve text extraction when scans are unreadable, empty, or poorly recognized.
- What file types are supported?
PDFs, common images (JPEG, PNG, HEIC, WebP, GIF), and plain text (.txt) — up to size limits per file.
- Large PDF uploads
Succeed with big PDFs by respecting size limits, splitting binders, and allowing extra processing time.
- Upload failures
Recover when Add Documents fails validation, stalls, or reports a failed batch.
Was this helpful?