“We know we're losing money on carrier adjustments — we just can't see where.”
All Case Studies
Integration & AutomationEcommerce Fulfillment / 3PL$10M–$50M RevenueProjected / target results
Recovering Leaked Carrier Spend for a Multi-Brand Fulfillment Agency
A multi-brand ecommerce fulfillment agency moves ~10K shipments a month across ~20 Shopify brands — but Shopify, ShipStation, the carrier invoices, and QuickBooks never talked to each other. Untraceable carrier adjustments quietly leaked ~$6K/month, and monthly client billing ate roughly a week of manual work. We built the integration and automation layer that reconciles every shipment and assembles billing in minutes. (Engagement is in UAT — figures below are projected/target.)
Systems we connected
Shopify (×20 brands)ShipStationUPSFedExMail InnovationsQuickBooks
01
The Company
A multi-brand ecommerce fulfillment agency that runs storefronts and shipping operations for roughly 20 consumer brands, processing on the order of 10,000 shipments per month. Each brand sells on Shopify; orders flow to ShipStation for fulfillment; parcels ship via UPS, FedEx, and Mail Innovations; and client billing is reconciled in QuickBooks. The systems each do their job — but nothing connects them, so the agency stitches exports together by hand to figure out what each brand actually owes.
Business Model
Multi-brand ecommerce fulfillment / 3PL
Brands Served
~20 Shopify storefronts
Shipment Volume
~10,000 / month
Carriers
UPS, FedEx, Mail Innovations
Accounting
QuickBooks
Connected Systems
Shopify, ShipStation, carrier invoices, QuickBooks
02
The Problem
Carrier invoices arrive full of after-the-fact adjustments — dimensional reweighs, surcharges, address corrections — that never make it back to the original order. With ~10K shipments a month across three carriers, those adjustments were impossible to trace by hand, and an estimated ~$6K/month of leaked spend went unbilled or unrecovered. On top of that, assembling each brand's monthly invoice meant manually cross-referencing ShipStation exports, carrier statements, and Shopify orders — roughly a week of work, every month, with errors baked in.
“Closing the books each month means a week of someone living in spreadsheets.”
“Every brand's bill is a manual reconciliation against three different carrier statements.”
03
What We Built
We built the custom integration pipeline that connects Shopify → ShipStation → carrier invoices (UPS, FedEx, Mail Innovations) → QuickBooks into one reconciled view of every shipment. On top of it sits a deterministic reconciliation and billing-automation layer: the matching and the math are done in auditable software, not by an AI. AI is used only for the ambiguous parts — parsing inconsistent carrier-invoice formats and summarizing exceptions for a human — and never to compute a charge.
Unified Order Pipeline
Connects each brand's Shopify storefront to ShipStation and ties every fulfilled parcel back to its originating order — one normalized record per shipment across all ~20 brands.
Carrier Invoice Ingestion
Ingests UPS, FedEx, and Mail Innovations invoices and normalizes their inconsistent formats into a single schema. AI assists only with ambiguous line parsing — it never decides what is owed.
Three-Way Reconciliation
Deterministically matches each shipment to its carrier charge and to the amount billed to the brand — auditable line by line, so every dollar ties out. The math is software, not a model.
Leakage Detection
Flags untraceable adjustments — reweighs, surcharges, address corrections — that previously slipped through, turning ~$6K/month of invisible leakage into a recoverable, line-itemized list (projected).
Automated Client Billing
Assembles each brand's monthly invoice from the reconciled data and pushes it into QuickBooks — collapsing roughly a week of manual cross-referencing into minutes (projected).
AI Exception Assistant
Summarizes the exceptions a human still needs to review — odd adjustments, unmatched parcels, format anomalies — in plain language, so the ops team triages in minutes instead of hours.
04
The Numbers
Projected outcomes from the UAT-stage build — modeled from the agency's own shipment and invoice data.
These figures are projected / target outcomes from a UAT-stage build, modeled from the client's own data — not yet realized in production.
~$6K/moCarrier leakage recovered (projected)
~$72K/yrAnnualized leakage recovery (projected)
1 week → minutesMonthly billing assembly (projected)
~10K/moShipments reconciled automatically
100%Deterministic, auditable reconciliation math
~20Brands billed from one pipeline
05
The Timeline
Not 6 months. Not 12 months. Weeks.
Week 0
Discovery Call
Mapped the order-to-invoice flow, the carrier mix, and where the manual billing time actually goes.
Week 1–2
Pipeline Prototype
Connected Shopify and ShipStation and ingested a month of real carrier invoices to prove the matching logic.
Week 2–3
Reconciliation Engine
Built the deterministic three-way match and validated it against a known billing period the team had closed by hand.
Week 3–5
Billing Automation
Automated per-brand invoice assembly and the QuickBooks push, with leakage flagged as a reviewable list.
Week 5–6
UAT
Agency is validating projected recovery and billing accuracy against live months before full cutover.
Next
Production Cutover
Move monthly close onto the automated pipeline and begin tracking realized recovery against projections.
06
The Reaction
“For the first time we can point at a specific shipment and a specific carrier adjustment and say exactly what it cost us. That alone changes the conversation with our carriers.”
Founder
Founder & CEO
“Billing used to be the week everyone dreaded. Watching it assemble itself from reconciled data — and tie out to the penny — is a different business.”
Operations Lead
Head of Operations
“The part that sold me is that the math is deterministic. The AI summarizes the weird stuff; it never decides what a brand owes.”
Controller
Controller
Two very different builds. The same missing layer.
Whether you need to see your business or run it without manual glue — let's talk about what one unified layer looks like on your data.