dc.codes
Commerce Ops Dashboard
Commerce Ops Dashboard
Scroll

Commerce Ops Dashboard

E-commerce operations SaaS, Hong Kong

Concept studyAn illustrative concept showing the kind of work we do, not a client engagement.

Services
Web app, Backend & API, Product design, AI integration
Tech
Next.js, TypeScript, PostgreSQL, Redis, Claude API
Region
Hong Kong · 2024

E-commerce operations teams were stitching orders, returns and stock levels together from spreadsheet exports, so a spike was usually noticed days after it mattered. We built a multi-tenant analytics dashboard on Next.js and PostgreSQL with live order, returns and inventory charts, role-based access per store and team, scheduled PDF and CSV exports, and an assistant that explains any spike in plain language and shows the queries behind its answer. We designed and shipped it in short iterations, using AI-assisted tooling for scaffolding, query tuning and test coverage while every change still went through human review.

Duration
6 months, ongoing
Engagement
Dedicated team
Platforms
Web · Admin console
Status
Live

What we built

01

Live order, returns and inventory charts with channel, region and warehouse filters

02

Multi-tenant data isolation with role-based access per store and team

03

Scheduled PDF and CSV exports delivered automatically to chosen roles

04

AI assistant that explains spikes and shows the SQL behind each answer

Approach

We started with one pilot tenant, rebuilding its weekly spreadsheet reports as dashboards so the ops managers could check every number against figures they already trusted. Automated tests verify tenant isolation on every release, and the assistant's answers are compared with hand-written queries before any new prompt version ships.

Overview for one tenant: live orders, returns by reason, low-stock alerts and scheduled exports.
01Overview for one tenant: live orders, returns by reason, low-stock alerts and scheduled exports.
Returns view: the assistant breaks an anomaly into its causes, with the queries behind it one click away.
02Returns view: the assistant breaks an anomaly into its causes, with the queries behind it one click away.
Contact Us