How to Import Scanned Receipts into Google Sheets
Turn a pile of scanned receipts into a clean Google Sheet. Extract data, export as a BOM-safe CSV, and import with one click. Every value is verifiable.
That stack of receipts isn't getting any smaller. Whether for expense reports or bookkeeping, the process is always the same: tedious, manual data entry into a spreadsheet. You type a vendor name, a date, a total, and then move to the next line, hoping you didn't make a typo. There's a more direct path from that scanned image to a clean row in Google Sheets.
See how it works on a real receipt below, without uploading anything.
Each value with a box carries a verified on-page location — bbox + 4-point vertices + match_ratio — on a 0–1000 normalized grid (0,0 top-left → 1000,1000 bottom-right), the same shape the live API returns. Hover a field to trace it back to the pixels it came from.
The flow is simple. First, you define the columns you need for your receipts—things like Vendor, Date, Total, and maybe some line items. This creates a structured sheet inside space-ocr. Then, you just upload your scanned images (JPEG, PNG, etc.). The service processes each one, placing the extracted data into the correct columns as new rows.
When you're ready, you export the entire collection as a single CSV file.
This isn't a complex integration. It's a clean, standard CSV file that Google Sheets understands perfectly. Because the file is encoded with a UTF-8 BOM, characters from any language—like the store name 'ライフ' (Life) from one of the receipts in the demo—import without issue. You just use the built-in File > Import function in Google Sheets.
What's different is the audit trail. You're not just getting a list of text values; you're getting data that is permanently tied to its source on the page.
To ensure every value is grounded in reality, space-ocr cross-references the output from its language model against the document's source. The LLM returns a text value along with hints about which words it used. The engine then performs a character-by-character matching of that value against the actual OCR symbols detected on the page.
This process generates a confidence score (match_ratio) for each field. A score of 0.85 or higher indicates a confident match. The result is a bounding box (xmin, ymin, xmax, ymax) with 0-1000 normalized coordinates, showing exactly where the data was found.
The pricing is straightforward. Each successful image processed costs $0.05. If a document fails to process for any reason, you are not charged. Your first 100 images each month are free, with no credit card required to get started.
- Create a Sheet in space-ocrDefine the columns you need for your data, such as Vendor, Date, and Total.
- Upload Your ScansDrag and drop your JPEG, PNG, or other receipt images onto your new sheet.
- Review and VerifyCheck the extracted data. Click any cell to see its origin on the original scan.
- Export to CSVFrom the sheet menu, download all the extracted rows as a single CSV file.
- Import into Google SheetsIn a new or existing Google Sheet, go to File > Import and upload the CSV file you downloaded.