AuditWedge
A margin-leak audit tool for cross-border e-commerce sellers
Supplier invoices are messy, FX markup is often invisible, and per-SKU profitability gets lost between three disconnected systems — the supplier invoice, the Shopify checkout, and the payout statement. AuditWedge reconciles all three at the SKU level to surface where margin is actually leaking.
Status: in active development, not yet validated with a paying customer. The extraction and reconciliation pipeline is built and passing internal accuracy testing against a synthetic invoice dataset — it has not yet been run against a real merchant's data or confirmed to catch a real discrepancy.
What's built so far
Invoice Extraction
Claude-based structured extraction of SKU, quantity, unit price, and currency from unstructured supplier invoices, with a confidence score on every field and no invented values.
Order↔Invoice Reconciliation
Joins extracted invoice line items to Shopify order data by exact SKU, with a title-similarity fallback for missing SKUs that explicitly flags unconfirmed variant matches for human review rather than guessing.
Margin Audit Report
Computes per-SKU contribution margin and flags supplier price drift, ranked by dollar impact, with every flagged leak citing its source invoice line and order.
Trust-First by Design
Every claim in the output carries a confidence score and a traceable source record; ambiguous cases are flagged for review, never silently resolved either way.
Not yet built
Ad-spend attribution (Meta/TikTok) and FX-leakage detection on supplier payouts are scoped but not implemented — the report explicitly labels these as unavailable rather than omitting them.
Built with Python, the Claude API, and Shopify's Admin GraphQL API. Pre-revenue, pre-validation.
Read the sample audit report