Designing a Universal OMS: End‑to‑End Order Management Architecture

Overall Design Goals (Four Core Objectives)

Unified order collection: aggregate orders from all sales channels into a single OMS.

Unified order standardization: convert heterogeneous channel formats into a common data model.

Unified order scheduling and distribution: automatically split orders and dispatch them to WMS (warehouse) and TMS (logistics).

Full‑lifecycle control: track every stage from order creation, payment, audit, picking, delivery, receipt, after‑sale, to financial reconciliation.

Complete Business Process

Channel order → OMS automatic pull/push → order standardization & cleaning → risk‑control validation → manual/automatic audit → split‑merge processing → warehouse WMS dispatch → logistics sync → user receipt → reverse after‑sale (refund/return) → financial reconciliation and archiving.

Layered Architecture & Four‑Level Functional Modules

1. Channel Integration Layer

1.1 Channel Connection Management

1.1.1 Channel Basic Configuration (record JD, Douyin, Mini‑Program, offline store, distributor info)

Channel Authorization (API keys, secret)

Price Isolation per channel

Inventory sync toggle per channel

1.2 Order Synchronization Tasks

Scheduled pull tasks (periodic order fetching)

Real‑time push listeners (callback handling)

Retry mechanism (3 automatic retries on timeout/network issues)

Synchronization logs for traceability

1.3 Channel‑Product Mapping

SKU binding between platform codes and internal SKU

Automatic specification conversion (e.g., "Red L" → standardized spec text)

Channel product disable control (auto block when product is off‑shelf)

Mapping exception alerts for unmatched codes

1.4 Channel Message Push Center

Order status callbacks to platforms (shipping info, cancellation, after‑sale status)

Failure‑retry queue for push timeouts

Platform‑side message notifications (order success, shipping reminders)

Activity sync (coupons, promotions) and push log query

2. Order Core Processing Layer

2.1 Order Standardization

Field standardization across channels

Address parsing into province/city/district/detail

Unified discount breakdown (platform coupons, store coupons, promotions)

Data completion for missing phone, remarks, etc.

2.2 Order De‑duplication

Channel order‑ID deduplication

Short‑time duplicate order detection (same account multiple orders within 1 minute)

Deduplication logs for audit

Manual recovery of falsely blocked orders

2.3 Order Classification

Normal spot‑goods orders

Pre‑sale orders (deposit & final payment separation)

Wholesale bulk orders (identified for distributors)

Virtual/gift‑card orders (no physical shipment)

2.4 Risk‑Control Center

Basic rules: address fraud, phone‑number abuse, purchase‑limit violations, inventory pre‑allocation checks

User‑behavior checks: high‑frequency ordering, high‑return users, cross‑region anomalies, blacklist matching

Payment checks: large‑order manual review, payment timeout cancellation, abnormal discount detection, illegal payment channel identification

2.5 Order Audit & Scheduling

Automatic audit for low‑risk small orders, pre‑sale completion, wholesale tiered auto‑approval

Manual audit workbench (list of risk‑blocked or large orders, one‑click release, rejection with reason, batch processing)

Split‑&‑Merge Engine (multi‑warehouse split, multi‑supplier split, same‑user merge, logging)

2.6 Fulfilment Dispatch

WMS dispatch: pick‑list generation, nearest‑warehouse strategy, retry on failure, status callbacks

TMS dispatch: automatic carrier matching, waybill creation, freight calculation, real‑time tracking pull

2.7 Forward Query Center

Full‑order search (multi‑criteria filtering, detail view, Excel export, custom tags)

Statistics dashboard (channel sales, order status distribution, hot‑selling items, exception summary)

3. Reverse After‑Sale Layer

3.1 After‑sale Order Integration

Automatic pull of platform refund/return requests

Manual entry for offline store returns

Binding after‑sale order to original order

Automatic classification (refund‑only, return‑refund, exchange, price‑protection)

3.2 After‑sale Risk Control

Time‑limit validation (no refund after after‑sale window)

Duplicate after‑sale detection

Malicious return identification (high‑frequency returns flagged for manual review)

Refund amount limit (cannot exceed original payment)

3.3 After‑sale Audit

Small‑amount automatic refunds

Manual workbench for large refunds

Refund amount adjustment by operations

Reject refund with reason notification

3.4 Return/Exchange Processing

Automatic push of return address to warehouse

Exchange product stock verification

Freight responsibility rule (merchant vs user)

Return tracking until user ships back

3.5 Reverse Inventory

Return‑goods inbound sync

Damaged‑goods marking and partial refund

Exchange new‑product dispatch

Reverse logistics tracking

3.6 Refund Finance Management

Original‑channel refund matching

Split refunds for a single order

Coupon reinstatement after return

Points/credits refund calculation

Refund ledger query, failure handling, export for accounting, loss statistics

4. Finance & System Support Layer

4.1 Financial Reconciliation Center

Automatic platform bill download

Order‑bill matching, discrepancy highlighting, manual discrepancy handling workbench

Commission deduction, gross‑margin calculation, multi‑dimensional financial reports, Excel export for ERP import

4.2 System Foundations

RBAC user & role management, data permission isolation, operation logs

Configuration Center (risk rules, split/merge policies, auto‑audit thresholds, API timeout/retry settings)

Task & Message Scheduling (periodic pull, cancel timeout, logistics sync, queue monitoring, retry, alert notifications)

System Log & Alert Center (full‑link order logs, anomaly alerts, operator traceability, monthly log archiving)

Implementation Steps (Four Phases)

Requirement gathering & basic configuration: map all sales channels, import SKUs, set up risk, split, auto‑audit rules, integrate WMS/TMS APIs.

Channel integration testing: API connectivity, end‑to‑end order, payment, delivery, after‑sale flow; simulate duplicate orders, stock shortage, large‑order checks.

Gray‑scale trial run: limited channel orders processed manually, verify reconciliation, discount split, amount calculations, collect operational feedback, adjust rules.

Full production & operation iteration: all channels live, enable alerting for missing/abnormal orders, monthly financial export, continuous feature iteration for new channels.

Common High‑Frequency Issues & Mitigation Strategies

Missing orders: dual sync (scheduled pull + real‑time callback), daily full reconciliation, alert thresholds, log replay.

Price chaos across channels: isolated SKU mapping, price‑change approval workflow, real‑time price validation, price‑change audit log.

Warehouse dispatch failures: delayed‑queue retry (3 attempts at 1/3/10 min), manual re‑dispatch workbench, warehouse heartbeat monitoring, inventory lock‑release on failure.

Multiple orders from same user causing high freight: custom merge engine rules (same user/address within 30 min auto‑merge), manual merge option, merge logs for audit.

After‑sale order mismatches: enforce original channel order binding, strict refund amount check.

Peak‑sale traffic overload: asynchronous processing via message queues, sharding tables by channel/date, off‑peak scheduled tasks, dynamic throttling of non‑core dashboards.

Unauthorized data access: RBAC isolation per channel/company, role‑based function limits, secondary verification for large refunds or batch cancellations, full audit trail.

Large reconciliation discrepancies: automatic fee breakdown, auto‑match platform bills, red‑flag differences, manual annotation, batch export for platform support.

Fraudulent bulk buying or coupon abuse: multi‑layer risk rules (phone, address, device, frequency), blacklist, manual review for high‑value discount orders, daily fraud statistics.

Duplicate push status causing repeated pick lists: globally unique idempotency token per order/status push, idempotency check discards repeats, idempotent operation logs, pre‑dispatch status verification.