I Built an Open-Source Alternative to Google Analytics + Hotjar

#analytics #heatmaps #automation #opensource

I was paying for Google Analytics (confusing), Hotjar ($80/mo for heatmaps), and a separate popup tool ($30/mo). Three dashboards, three scripts on my site, and nothing talked to each other.

So I built Seentics — one open-source tool that replaces all three.

What Seentics Does

Analytics — Real-time visitors, pageviews, bounce rate, session duration, traffic sources, UTM tracking, countries, browsers, devices. Like Plausible but with way more features.

Heatmaps — See where users click and how far they scroll. Supports desktop, tablet, and mobile viewports. Works on any page without code changes.

Session Replays — Watch full user sessions to find UX issues and bugs. Stored in S3-compatible storage with automatic PII masking.

Funnels — Build multi-step conversion funnels. See exactly where users drop off in your signup, checkout, or onboarding flow.

Goal Tracking — Track page visit goals, custom events with seentics.track('signup'), or auto-track clicks on CSS selectors like #buy-btn.

Behavioral Automations — This is the feature nobody else has in the privacy-first space. Trigger popups, banners, webhooks, or custom JavaScript based on real-time visitor behavior — exit intent, scroll depth, time on page.

The Tech Stack

Layer	Technology
Backend API	Go 1.24 (Gin)
Frontend	Next.js 14, Tailwind CSS, shadcn/ui
Analytics DB	ClickHouse
Metadata DB	PostgreSQL 15
Object Storage	S3-compatible (MinIO)
Caching	Embedded Go cache

Why Go + ClickHouse?

Go handles concurrent HTTP requests extremely well — the event ingestion endpoint needs to handle thousands of requests per second with minimal latency. The tracking script sends a POST request, Go buffers it in memory, and batch-inserts into ClickHouse every second or every 1000 events (whichever comes first). Response time to the browser: under 5ms.

ClickHouse is a columnar database built for analytical queries. A query like "show me pageviews by country for the last 30 days" that would take 10+ seconds on PostgreSQL runs in 50ms on ClickHouse — even with millions of rows.

PostgreSQL handles everything that's not time-series: user accounts, site configurations, goals, automation rules.

The Tracking Script

Under 1KB. Compare that to GA4 (45KB+) or Hotjar (100KB+). No cookies, no fingerprinting, no PII collection.

Setup is one line in your HTML:

<script
  async
  src="https://seentics.com/trackers/seentics.js"
  data-site-id="YOUR_SITE_ID"
></script>

Custom event tracking:

// Basic event
seentics.track('signup_click')

// With properties
seentics.track('purchase', { value: 49.99, plan: 'pro' })

How It Compares

Feature	Seentics	Plausible	Hotjar	PostHog	GA4
Real-time analytics	Yes	Yes	No	Yes	Yes
Heatmaps	Yes	No	Yes	No	No
Session replays	Yes	No	Yes	Yes	No
Funnels	Yes	Yes	Yes	Yes	Yes
Behavioral automations	Yes	No	No	No	No
Privacy-first (no cookies)	Yes	Yes	No	Yes	No
Open source	Yes	Yes	No	Yes	No
Self-hostable	Yes	Yes	No	Yes	No
Script size	<1KB	<1KB	~100KB	~25KB	~45KB

Self-Hosting

Requires Docker. That's it.

git clone https://github.com/Seentics/seentics.git
cd seentics
docker compose up -d --build

Open http://localhost:3000 and you have a full analytics platform running.

Architecture

Browser --> Tracking Script --> Go Backend API (:3002)
                                     |
                   +-----------------+-----------------+
                   |                 |                 |
              ClickHouse        PostgreSQL           MinIO
              (events)          (metadata)         (replays)

The tracking script is served by the Go backend and sends events directly to it. Events are buffered in memory and batch-inserted into ClickHouse. Session replay data goes to S3-compatible storage (MinIO for self-hosted, any S3 for cloud). PostgreSQL stores user accounts, site configs, goals, and automation rules.