OCR isn't reading my receipts correctly

Nothing is being detected at all

• Check the image isn't too dark or blurry. OCR needs decent contrast between text and background — a photo taken in dim light is the single most common reason for a blank read.
• Check the upload actually went through. Try re-uploading; a partial or corrupted upload tends to fail silently rather than produce an obviously bad read.
• Give it a moment. OCR runs on your device rather than a server, so how long it takes depends on your device's speed. A few seconds for a full-page receipt is completely normal.

Numbers are being read wrong

This is almost always a photo-quality issue rather than an app bug. Check your photo against these:

The total's right but the tax line is wrong

Tax lines are usually the smallest text on a receipt, and the easiest to misread as a result. It's worth glancing at the tax field before you save, since it's a two-second check that feeds directly into your input tax credit claim.

The vendor name comes out garbled

Store logos and stylized headers are genuinely harder for OCR than plain printed text. That's a known limitation, not a data problem. Just correct the vendor field by hand; it takes a second, and if it's a vendor you see often, pair it with a vendor rule so future receipts still categorize correctly even when the OCR read on the name itself isn't perfect.

A note on privacy

Since OCR runs locally in your browser, there's no server-side copy to inspect or debug remotely, which is exactly the privacy trade-off GBooks Air is built around. If one type of receipt consistently reads poorly, email a description of it (not the receipt itself, to keep it out of your inbox) to support@gshfinancial.com.