Case Study · 12 · Real-Time Traffic Intelligence

O Trânsito Agora

Everything happening on Portuguese roads right now — traffic jams, accidents, active fires near roads, weather warnings, road closures, and transit disruptions — in one fast, clean, mobile-first map. Built on the same public data the authorities use, presented in a way that actually helps you decide when to leave, which route to take, and what to avoid. No app to download. No signup. Free.

Visit transitoagora.pt

Project: O Trânsito Agora
Sector: Civic data · Mobility · Real-time
Role: Design + build · Solo
Status: Live

01 — The Brief

Six data sources. Zero unified view.

Portuguese drivers check traffic before every commute, but no single source shows the full picture. Google Maps shows congestion but ignores fires, ProCiv incidents, and transit alerts. Fogos.pt covers fires only. IP's official traffic page has terrible UX and no mobile experience. IPMA has weather warnings. Carris has transit alerts. Each source lives in its own silo.

The gap isn't data — it's aggregation. A commuter in Lisbon needs to cross-reference four tabs to answer a simple question: 'should I leave now or wait 30 minutes?' The brief: one map, all layers, real-time, mobile-first.

02 — The Approach

Unified incident schema. Map-first. Content auto-generated.

The data pipeline is the product. A Railway cron service polls six sources at their natural cadences — TomTom traffic every 60 seconds (cached), ProCiv every 5 minutes, ICNF fires every 10 minutes, IPMA hourly. Every event gets normalised into a unified Incident schema with type, severity, coordinates, source, and timestamp. Postgres with PostGIS handles spatial queries.

The front-end is map-first: MapLibre GL with TomTom vector tiles for the traffic layer, custom markers for incidents, fire perimeters, and transit alerts. Toggle layers on/off. Filter by severity. Geo-search by district. The hero loads the map viewport, not a marketing splash.

Content is auto-generated from the data. Daily traffic reports per district, weekly trend summaries, incident archives — all programmatic. Every district and incident type gets its own SSG page for the long-tail of 'trânsito lisboa hoje' queries.

03 — The Build

PostGIS, WebSocket updates, email alerts.

Stack: Next.js 15 + Postgres with PostGIS + Drizzle on Railway. A separate cron service ingests from TomTom, ProCiv, ICNF, IPMA, Carris, and CP. Each source adapter normalises into the shared Incident schema and upserts with deduplication by source + external ID.

Real-time updates push to connected clients via Server-Sent Events. The map refreshes incident markers without full page reload. Historical data feeds the auto-generated daily reports and trend pages.

Email alerts use district-level subscriptions: 'email me when a severity-3+ incident happens in Lisboa'. Double opt-in, per-district unsubscribe, anonymous beyond email + district preference. Resend handles delivery.

SEO surface: every district gets a daily report page, every incident type gets an archive page, and the homepage targets 'trânsito portugal' and 'acidentes portugal hoje'. ~400 SSG pages at launch, growing daily with auto-generated reports.

Key Features

What makes the product usable.

Unified real-time map

Traffic flow, accidents, fires, weather warnings, transit disruptions — all on one MapLibre GL map with toggleable layers and severity filters.

Six official data sources

TomTom traffic, ProCiv incidents, ICNF fires, IPMA weather, Carris transit, CP rail — polled at each source's natural cadence and normalised into one schema.

Auto-generated daily reports

Every district gets a daily traffic report page with incident counts, severity breakdown, and trend comparison to the previous week. Zero editorial effort.

District-level email alerts

'Alert me when something serious happens on roads in my district'. Double opt-in, per-district unsubscribe, anonymous.

Mobile-first dark map UI

Designed to be opened on a phone before leaving home. Map loads first. Layers toggle with one tap. District filter in the header.

~400+ SSG pages for SEO

District daily reports, incident archives, trend pages — all targeting the long-tail of 'trânsito <distrito> hoje' and 'acidentes <distrito>' queries.

04 — The Stack

Every piece picked for a reason.

Next.js 15

App Router; SSG for report pages, RSC for fresh map data, API routes for real-time endpoints

Postgres + PostGIS + Drizzle

Spatial queries for incidents near point, time-series for trends, unified incident store

TomTom Traffic API

Real-time flow + incidents; free tier covers 50K tiles/day

MapLibre GL JS

Vector tiles, custom styling, toggleable layers, free and open-source

ProCiv + ICNF + IPMA

Official Portuguese public data for incidents, fires, weather

Server-Sent Events

Push incident updates to connected map clients without WebSocket overhead

Resend

District-level email alerts with double opt-in

Railway

Web service + Postgres + cron workers in one project

Plausible Analytics

GDPR-clean, no cookie banner

05 — Results

One map that replaces four tabs.

O Trânsito Agora launched aggregating six sources into a unified real-time view. The map loads in under 2 seconds on mobile, shows every active incident across Portugal, and lets users toggle layers to focus on what matters — fires during summer, transit disruptions during strikes, accidents on their commute route.

Auto-generated daily reports create a growing SEO surface without editorial effort. Each district report targets high-intent queries and builds internal links naturally. The SSE-powered live updates keep the map current without page refreshes.

The product fills a gap no one else covers: aggregated, real-time, all-source traffic intelligence for Portugal in one free, mobile-first interface.

Data sources: 6
Districts covered: 20
Map load time: < 2s
Incident refresh: ~60s

06 — Takeaways

What we learned aggregating real-time public data.

01
The value is in aggregation, not in any single source. No one source tells the full story.
02
Poll each source at its natural cadence. Over-polling wastes quota; under-polling makes the map feel stale.
03
Auto-generate content from structured data. Daily reports write themselves and compound SEO value.
04
Map-first, not content-first. The user wants to see, not to read.
05
SSE beats WebSocket for one-directional updates. Simpler, reconnects automatically, works through proxies.

07 — What's next

Have fragmented public data that needs a unified view? Same approach.

Start a conversation See more case studies

O Trânsito Agora

Visit transitoagora.pt

Six data sources. Zero unified view.

Unified incident schema. Map-first. Content auto-generated.

PostGIS, WebSocket updates, email alerts.

One map that replaces four tabs.

The product fills a gap no one else covers: aggregated, real-time, all-source traffic intelligence for Portugal in one free, mobile-first interface.

What we learned aggregating real-time public data.

The value is in aggregation, not in any single source. No one source tells the full story.

Poll each source at its natural cadence. Over-polling wastes quota; under-polling makes the map feel stale.

Auto-generate content from structured data. Daily reports write themselves and compound SEO value.

Map-first, not content-first. The user wants to see, not to read.

SSE beats WebSocket for one-directional updates. Simpler, reconnects automatically, works through proxies.