QA & TestingNext.jsPerformance

Testing Smart Tour

How DevNex built, tested, and launched a multi-role, AI-integrated travel platform for northern Pakistan with zero critical defects and 95ms average response time under load.

Project

Smart Tour - AI Travel Platform

Live URL

smarttour-v2.vercel.app

Tech Stack

Next.js 14, Supabase, Tailwind, AI

Delivered By

DevNex Innovation

Executive Summary

100%

Black Box Pass Rate

95.63ms

Avg Response Time

714

Load Test Requests

DevNex designed, built, and QA-tested Smart Tour a full stack AI-powered travel planning platform for northern Pakistan tourism across three comprehensive testing phases. Black Box Testing validated all 8 core user journeys with a 100% pass rate. API Testing via Postman confirmed all 4 production endpoints returned HTTP 200 OK with sub 340ms response times. Load Testing via k6 demonstrated the system handles 10 concurrent users at 95.63ms average response time well below the industry benchmark of 200ms.

Project Context

Smart Tour is a full stack SaaS web application purpose-built for northern Pakistan tourism. The platform serves three distinct user roles through purpose-built portals: Traveler, Tour Company, and Administrator. It integrates real-time AI, live weather data, safety mapping, and a complete booking management system.

Goals & KPIs

Testing Phase	KPI Target	Category
Black Box Testing	100% pass rate across all core user journeys	Functional
API Testing	HTTP 200 OK on all endpoints, response time under 500ms	Reliability
Load Testing	Avg response time under 200ms with 10 concurrent users	Performance
Security	Protected routes must reject unauthenticated access correctly	Security

Real Testing Challenges & Solutions

Challenge 1: Access Control

Ensuring traveler, company, and admin roles could not access each other's dashboards.

✓ Solution: Supabase RLS policies with a SECURITY DEFINER helper.

Challenge 2: Live APIs

Validating OpenWeatherMap, Mapbox GL JS, and Hugging Face in production.

✓ Solution: Isolated per-integration test cases confirming stable API health.

Challenge 3: Middleware Load

Simulating 10 VUs risked exposing session leakage across protected routes.

✓ Solution: k6 multi-URL load test confirmed consistent HTTP 302 redirects safely.

Challenge 4: Production Only

No staging environment meant testing against the live Vercel deployment.

✓ Solution: Pre-test production hardening (rotating keys, sanitizing environment).

QA Strategy Implementation

Phase 1: Black Box Testing

Validate every critical end-user journey simulating real user behavior from registration through core features.

8 test cases across all three roles.
Manual browser testing on Windows 11 against live deployment.
Result: 8/8 passed (100% functional pass rate).

Phase 2: API Testing (Postman)

Verify HTTP endpoint behavior independently of the frontend UI.

4 endpoints tested (/tours, /login, /api/weather).
All 4 returned HTTP 200 OK with sub 340ms response times.
Custom weather endpoint delivered data in just 265ms.

Phase 3: Load Testing (k6)

Simulate concurrent traffic to identify performance bottlenecks and stability issues.

10 Virtual Users (VUs) running for 30 seconds.
714 total requests processed with zero server crashes.
95.63ms average response time achieved.

Key Takeaways

Test in production carefully: Harden the production build completely before the test window begins. DevNex rotated credentials before sending a single test request.
Multi-role platforms need multi-phase testing: Black box testing catches functional gaps. API testing catches routing issues. Load testing catches authentication concurrency behavior.
Performance is a feature you measure: 95.63ms average response time did not happen by accident. It is the result of optimized Supabase queries and Next.js App Router architecture.

Need a platform built right and tested to prove it?

DevNex builds full-stack AI-powered SaaS platforms with end-to-end QA baked into the development process.

Book a Free Call