# diy-analytics

> diy-analytics is a free, open-source, self-hosted web analytics platform. You deploy it to your own infrastructure, connect your own PostgreSQL database, and own 100% of your visitor data — nothing is sent to a third party.

diy-analytics is a privacy-first, cookie-free alternative to Google Analytics, Plausible, and Umami. It is MIT-licensed, built with Next.js (App Router), Drizzle ORM, and PostgreSQL, and ships as a single deployable app (e.g. to Vercel).

## Key Facts

- **Name:** diy-analytics (also known as DIY Analytics)
- **License:** MIT (free, open source, self-host forever)
- **Stack:** Next.js (App Router), TypeScript, Tailwind CSS, PostgreSQL via Drizzle ORM
- **Tracking Script:** Under 2 KB, async/defer, zero performance penalty, privacy-friendly
- **Deployment Time:** Under 5 minutes on Vercel with Neon, Supabase, or any PostgreSQL 14+ database
- **Data Ownership:** 100% self-hosted — telemetry lives exclusively in your Postgres database
- **Compliance:** Cookie-free, GDPR, CCPA, and PECR compliant by design without requiring cookie consent banners
- **AI & MCP Ready:** Built-in Model Context Protocol (MCP) server for Cursor, Claude Desktop, and LLM tool calling

## Core Features

- **Real-Time Dashboard:** Instant traffic stats, top pages, referrers, UTM campaigns, geolocations, devices, and browser breakdowns.
- **Funnels & User Journeys:** Multi-step funnel visualization and conversion drop-off analysis.
- **Retention Cohorts:** Weekly visitor retention matrix and user lifetime activity analysis.
- **Error Tracking & Diagnostics:** Client-side error monitoring with source-map resolved stack traces, breadcrumbs, and regression alerts.
- **Ad-Hoc Session Explore:** Multi-parameter query builder to segment users by attributes, visits, and custom events.
- **Visitor Segmentation:** Automated behavioral buckets (Active & frequent, Active & occasional, Lapsing, Dormant).
- **Core Web Vitals:** First Contentful Paint (FCP), Largest Contentful Paint (LCP), Cumulative Layout Shift (CLS), and Interaction to Next Paint (INP).
- **Alerts & Notifications:** Threshold and anomaly alerts dispatched to Slack, Discord, webhooks, or PagerDuty.
- **Public & Shared Dashboards:** Password-free or tokenized read-only links for clients and public metrics.
- **AI Assistant & MCP Server:** Native MCP endpoint (`/api/mcp`) allowing AI assistants (Claude, Cursor, custom agents) to query metrics, cohorts, and errors in natural language.

## Comparisons

- **diy-analytics vs. Google Analytics:** diy-analytics is 100% self-hosted, stores data in your own Postgres database, requires zero tracking cookies, needs no cookie banners, and includes built-in error tracking and an AI copilot. Google Analytics is hosted by Google, uses cookies, shares data, and lacks self-hosting and error tracking.
- **diy-analytics vs. Plausible:** diy-analytics is completely free and open source, runs on your own infrastructure, and includes full funnel analysis, retention cohorts, source-mapped error tracking, and an MCP server. Plausible charges monthly subscriptions for hosted services and lacks error tracking.
- **diy-analytics vs. Umami:** diy-analytics provides multi-step conversion funnels, retention matrices, error telemetry, and an MCP AI integration, whereas Umami focuses only on pageviews and basic events without funnels or error logging.

## Official Resources

- Website: https://analytics.sagnik.fun/
- Documentation: https://analytics.sagnik.fun/docs
- Getting Started: https://analytics.sagnik.fun/docs/getting-started
- FAQs: https://analytics.sagnik.fun/faqs
- Changelog: https://analytics.sagnik.fun/changelog
- GitHub Repository: https://github.com/heysagnik/diy-analytics

## Author & Attribution

Developed by Sagnik Sahoo (@heysagnik). Licensed under MIT.