Expense reports die a slow death by data entry. A shoebox of crumpled receipts becomes an evening of squinting at faded thermal paper and typing totals into a spreadsheet one cell at a time. OCR flips that: photograph a receipt, extract the merchant, date, and total, and drop it into a sheet that does the math for you.
What OCR actually pulls off a receipt
A receipt is a small, dense document with a predictable shape: a merchant name up top, a list of line items in the middle, taxes and a total at the bottom. OCR reads those printed characters and hands them back as text you can sort and sum. The most useful fields to capture are the merchant, the date, the subtotal, the tax, and the grand total, which are exactly the columns an expense sheet needs.
For a quick total you only need plain text, but for a real expense log you want structure, which is where extracting straight into a spreadsheet earns its keep.
Reality check: thermal paper is hard
Be honest about the input. Our engine is Tesseract-based, strong on clean printed text and good on screenshots and scans, but receipts are often the worst-case image: faded thermal ink, crinkles, and a glossy surface that loves to catch glare. Fresh receipts read well; a six-month-old gas receipt that's been through the wash will not.
You'll get the best results when you:
- Shoot the receipt fresh, before the thermal ink fades.
- Flatten it against a dark, matte surface so the edges and text stand out.
- Light it evenly to kill the glare strip that thermal paper produces.
Step by step: receipt to spreadsheet row
- Photograph the receipt flat. Lay it on a dark surface, smooth out the creases, and shoot straight down with even light.
- Crop to the paper. Remove the table, your hand, and any background so the recognizer sees only the receipt.
- Choose your output. For a single total, image to text gives you the figures as plain text. For a structured log, use image to excel to pull line items into cells.
- Check the numbers. Verify the date, total, and tax against the paper; a misread digit in a money field is the one error you can't afford.
- Build your sheet. Drop each receipt into its own row and let formulas total the column.
Why a spreadsheet beats a wall of text
A total you can't sum is just a number you'll retype later. Pulling receipts into image to excel keeps the structure: merchant in one column, date in the next, amount in another. Then a single SUM formula gives you the report total, and you can sort by date or filter by category without touching a calculator.
If you process expenses in waves, photograph a whole stack first and run them through batch image to text in one pass, then paste the lot into your sheet. The same capture discipline that helps here is the discipline you'll use for business cards and the invoice workflows below.
Handling crumpled or faded receipts
If a receipt comes back garbled, the cause is the surface, not the software. Smooth it under a flat object before shooting, and try side lighting to lift faint thermal text off the page. When the numbers are genuinely unreadable, type those few fields by hand, you're still saving time on every clean receipt in the pile.
From receipts to a real expense process
Receipts and invoices are cousins. Once you're comfortable extracting receipt data, the same approach scales to vendor bills, which is exactly what our guide on automating invoice data entry walks through. The capture-and-extract habit also pairs with how to digitize handwritten notes if you keep handwritten expense logs.
Keeping expense data accurate and private
Money fields demand a second look. Always reconcile the extracted total against the paper before you file a claim, since a transposed digit is easy to miss and hard to explain. Treat receipts as sensitive too: they can reveal where you were and what you bought. Files you upload are processed for conversion and not kept around, but it's still good practice to delete local copies once your report is filed.
Frequently asked questions
Can OCR read faded thermal receipts?
It can read clearer ones well, but heavily faded thermal paper is genuinely hard for any OCR engine. Photograph receipts while they're fresh, flatten them on a dark surface, and use even side lighting to lift faint text. Expect to retype the occasional unreadable total by hand.
Should I extract to text or straight to Excel?
For a single total, plain image to text is fine. For an expense log you'll sum and sort, use image to excel so each field lands in its own cell and formulas can total the column automatically.
How do I process a big stack of receipts at once?
Photograph them all first, then upload the batch to batch image to text. You convert the whole stack in one pass and paste the results into your spreadsheet, instead of doing each receipt individually.
How accurate are the totals I get back?
On clean, well-lit receipts, very accurate, but never file a claim without checking. A single misread digit in a money field can throw off a whole report, so reconcile each extracted total against the original receipt before submitting.
Turn your next pile of receipts into a tidy expense sheet. Start with image to excel and pull the totals straight into spreadsheet rows.