Extract Receipt Data from Images with Make.com

Phone photos are the default expense artifact. PDFs from cashiers are optional; camera rolls are not. This scenario accepts JPEG/PNG (and WebP) through a Make webhook or Google Forms file upload, runs Cryvis Receipt extraction, and writes a clean row to Google Sheets.

Hub index: Make.com. PDF-heavy Drive folders: extract receipt data. Field-level Sheets design: receipt to Google Sheets.

Workflow

Two triggers share the same Cryvis + Sheets tail. Pick one.

Option A — Custom webhook (mobile / app)

[Webhooks: Custom Webhook]
   multipart or JSON+file URL
        |
        v
[HTTP: Get File]  (if URL-based)
        |
        v
[HTTP: Make a Request]  Cryvis Receipt
        |
        v
[Google Sheets: Add a Row]

Option B — Google Forms (field teams)

[Google Forms: Watch Responses]
        |
        v
[Google Drive: Download a File]  (form upload)
        |
        v
[Filter: image/* or pdf]
        |
        v
[HTTP: Make a Request]  Cryvis Receipt
        |
        v
[Google Sheets: Add a Row]
   + optional Slack ping

Modules

Google Forms tip: use a File upload question limited to images, required. Add short text fields for Employee and Project code — Cryvis will not invent those from the photo.

Cryvis HTTP configuration

MIME support for this path: jpeg, png, webp (plus pdf if you allow it on the form). Credits: 1 per image. A webhook that sends three images as three separate scenario runs costs three credits; a single request with multiple file parts returns results[] and still bills per image.

Docs: extractReceipt. Product page: apis/receipt.

Webhook payload patterns

Preferred: multipart webhook where Make receives the file directly and maps it into the Cryvis multipart field.

Alternate: JSON body with file_url + employee_id. Then:

1. HTTP GET the URL (signed S3/GCS link).
2. Pass binary into Cryvis.

Do not base64-embed large photos inside JSON if you can avoid it — Make payload size limits will bite on multi-megapixel receipts.

Response fields for image receipts

Image receipts often have weaker line_items than digital PDFs. Still map:

data.merchant.name
data.transaction_date
data.transaction_time
data.total_amount
data.tax_total
data.currency
data.receipt_number
data.receipt_type
data.location.city
data.payment.method
data.amount_paid

If transaction_date is null (cropped header), keep the Sheets row and set Needs Review = true via a Filter on empty date.

Response mapping to Sheets

Keep Photo URL clickable. Auditors open the image faster than they reopen the form response.

Destination specifics

Sheets

Use a tab named Mobile Receipts separate from PDF pipeline tabs. Mobile photos have higher Needs Review rates; mixing them with clean PDF extractions rows skews ops dashboards.

Conditional formatting: highlight rows where Total is empty or Needs Review is true.

Webhook consumers

Return a simple JSON response from the webhook scenario so the mobile client can show success:

{ "ok": true, "merchant": "North Cafe", "total": 28.4 }

Map those from Cryvis data in the Webhook response module.

Image quality edge cases

• Glare / shadow: totals may still extract; merchant name may not. Route to Needs Review when merchant.name is empty but total_amount exists.
• Multi-receipt collage: one photo of two paper receipts confuses line boundaries. Instruct users: one receipt per frame.
• HEIC from iPhone: Make/Drive may convert or reject. Prefer Forms settings that convert to JPEG, or add a Converter module before Cryvis. Cryvis accepts jpeg/png/webp/pdf — not HEIC as a first-class MIME in this integration.
• Screenshots of e-receipts: usually fine; treat as PNG.
• Rotated images: Cryvis handles common rotations; extreme skew still benefits from a retake.
• Low resolution zooms: cropping too tightly can drop the merchant header while leaving the total barcode area. Prefer full-ticket framing with a small margin.
• Flash blowout on thermal paper: white-hot highlights erase ink. Ask field staff to disable flash and use ambient light.

Pre-flight checks in Make

Before calling Cryvis, a cheap Filter saves credits:

1. MIME allowlist passes.
2. File size between ~20 KB and 10 MB (empty stubs and huge panoramas both fail usefully).
3. Optional: filename does not match IMG_0000 placeholders from failed camera apps.

If your mobile client can compute blur scores, reject client-side first. Otherwise rely on Needs Review when total_amount is null.

For Gmail attachment flows instead of camera uploads, use automate expense tracking.

Errors

Never retry unsupported MIME errors — they will not heal.

Test plan

1. Submit a sharp JPEG of a grocery receipt through Forms.
2. Verify Drive download + Cryvis success.
3. Confirm Sheets row: merchant, date, total, photo link.
4. Submit a pitch-black photo; confirm Needs Review path.
5. Hit the webhook with a PNG screenshot of an e-receipt.
6. Submit HEIC if your iOS device defaults to it — document whether conversion is required.

Mapping employee metadata

Cryvis extracts what is on the receipt. It does not know who submitted the photo. Always carry Form/webhook fields alongside data.*:

If client-reported amount and data.total_amount differ by more than your threshold (commonly 2–5%), set Needs Review = true and optionally Slack the submitter. That catches wrong-photo uploads without blocking the entire pipeline.

Production checklist

• Forms: limit file size (e.g. 10 MB) to protect Make operations.
• Webhook: authenticate with a shared secret query param or Make’s IP allowlist behind your API gateway.
• Rate: mobile offsites can spike; monitor Cryvis credits (1/image).
• Privacy: receipt photos can include personal card last4 in payment — restrict Sheets ACL.
• Pair with Slack approval if policy requires it (employee expense workflow).
• Publish a one-page “how to photograph a receipt” guide linked from the Form description.

Related guides

• Convert receipts to JSON (HTTP deep dive)
• Extract receipt data (Drive)
• Invoice extraction on Make.com
• Make.com hub

CTA

Camera roll to structured expense data is a single multipart call. Start with the Cryvis Receipt API and the extractReceipt reference — Bearer token, file part, JSON back with merchant and totals.