QA Lite — how it works.
QA Lite runs structured checks on parser or vendor JSON outputs. Submit 25–50 redacted samples and get a benchmark report with accept / review / reject verdicts, reason codes, and a buyer-critical issue rate.
It is designed for teams that already have receipt data flowing downstream and want to know whether unsafe outputs are reaching payout, fraud review, rebate, loyalty, or analytics workflows — without sending receipt images or OCR text for the first benchmark.
Field presence, math, structure, and buyer-critical flags.
QA Lite runs a defined set of checks on every JSON output in your batch. No configuration required — checks apply automatically to every sample.
- Required fields checked for presence and basic format (total, line_items, merchant, date)
- Math reconciliation: line item sum vs. subtotal; subtotal + tax vs. total
- Buyer-critical flags: missing total, zero-item receipt, structural failures that could affect payout
Accept, review, or reject — with reason codes for every flag.
Every sample in the batch gets a verdict. Math mismatches, missing fields, and structural failures each produce a reason code. The report shows your buyer-critical issue rate and includes redacted examples.