Custom Internal Dashboards & Operational Tools

Project Overview

Internal teams were relying on spreadsheets, manual reports, and siloed tools with no single place to see operations, sales, or support KPIs. The goal was to centralize data, surface key metrics in one place, and enable self-serve reporting so stakeholders could answer their own questions without waiting on engineering or data teams.

Timeline: 3 months (discovery, build, rollout). Role: Lead developer — requirements gathering with departments, data pipeline design, dashboard UI (React), and role-based access. Tech stack: React, Chart.js / Recharts for visualization, REST and GraphQL APIs for data, PostgreSQL for app and reporting DB, role-based auth (JWT + permission flags), and optional WebSocket/polling for real-time updates.

The Challenge

Data lived in fragmented sources across departments — CRM, ticketing, billing, and internal APIs — with no single source of truth. Manual reporting was consuming hours every week, and stakeholders had different access levels and data needs: admins needed full visibility, managers needed team-only views, and some users only needed read-only summary dashboards. We had to design a modular dashboard builder that could connect to multiple data sources, respect roles, and stay performant with large datasets.

Flow 1 — Dashboard Data Pipeline

From data sources through processing to the frontend.

Dashboard Data Pipeline Flow

Data Sources Processing Layer Frontend

REST APIs PostgreSQL CSV Uploads Webhooks

API Gateway
Middleware Data Normalizer Cache (Redis) WebSocket Server

React Dashboard Chart Components KPI Cards Live Updates

Flow 2 — User Role & Permissions

Auth and role-based access to dashboards.

User Role & Permissions Flow

User Login Auth Service Token Issued

Role Check
Admin / Manager / Viewer

Full Access
All Dashboards Team Data Only
Dept. Dashboards Read Only
Summary View

Flow 3 — Widget Build & Deploy

From widget definition to live deployment.

Widget Build & Deploy Flow

Define Data Source Configure Widget Type

Map Fields & Filters

Preview in Sandbox Adjustments?
loop back

Publish to Dashboard

Assign to Role(s) Live for Users

Solution Walkthrough

Stage 1

Modular dashboard builder

Designed a layout engine where each dashboard is a grid of configurable widgets. Admins can add, remove, and resize widgets (KPI cards, charts, tables) and save layouts per role or department.

Stage 2

Data source connectors

Built connectors for REST/GraphQL APIs, direct PostgreSQL queries (with safe parameterization), and CSV uploads. Each widget declares its data source and refresh interval; the backend normalizes and caches where needed.

Stage 3

Role-based views

Implemented Admin (all dashboards and config), Manager (team or department dashboards only), and Viewer (read-only summary). Permissions are enforced server-side; the UI only renders what the token allows.

Stage 4

Real-time updates

Added optional WebSocket subscriptions for live metrics (e.g. support queue depth) and fallback polling for other widgets so the UI stays current without full page reloads.

Stage 5

Export, filter, drill-down

Each chart and table supports date-range and dimension filters; tables have search, sort, and pagination. Export to CSV is available for permitted roles so teams can pull data into their own workflows.

Feature Highlights

Wireframe-style preview of key UI patterns.

KPI card row

Total Users

12,847

↑ 4.2% vs last week

Revenue Today

$24.2k

↑ 12% vs yesterday

Open Tickets

89

↓ 8 from yesterday

Conversion Rate

3.8%

→ flat

Line chart — Weekly Active Users

Weekly Active Users

Data table (search, sort, pagination)

ID	Name	Department	Status	Last active
1001	Project Alpha	Ops	Active	2h ago
1002	Project Beta	Sales	Pending	5h ago
1003	Project Gamma	Support	Closed	1d ago

Sidebar navigation with role indicator

Results & Impact

Before

Spreadsheets and manual reports across departments
No single source of truth for operations or sales KPIs
Hours spent weekly on recurring report builds
Stakeholders with different access needs and no unified view

After

Centralized dashboards with live data from APIs and DB
~60% reduction in time spent on routine reporting
5 departments using self-serve dashboards daily
Role-based access (Admin / Manager / Viewer) with clear boundaries

Metric	Before	After
Weekly reporting time (avg)	~8 hrs	~3 hrs
Departments with dedicated dashboards	0	5
Self-serve report requests to IT	High	Low

"We finally have one place to see how we're doing. The dashboards replaced our weekly spreadsheet ritual and the team actually uses them every day."

— Operations lead (client)

Lessons Learned

• Designing for non-technical users: Default views and clear labels mattered more than power features. We simplified the first release and added advanced filters in v2 once adoption was solid.
• Performance with large datasets: Server-side aggregation and pagination were mandatory. We added caching for heavy queries and let the UI request only the date range and dimensions needed.
• Flexibility vs. simplicity: We resisted building a full drag-and-drop report builder in v1. Instead we shipped a fixed set of widget types and layouts; that constraint kept the UX predictable and supportable.