Matomo

Matomo's architecture reflects its age: a PHP application with MySQL/MariaDB storage, designed when analytics meant counting pageviews in server logs. It's been modernised — ClickHouse support, API improvements, a tag manager — but the core data model and processing pipeline predate the event-based paradigm that defines modern analytics. This creates friction for teams trying to implement behavioural analytics patterns on a pageview-era foundation.

The self-hosted value proposition is Matomo's most defensible advantage. In post-Schrems II Europe, where every US data transfer requires individual assessment and supplementary measures, running analytics on your own EU infrastructure eliminates an entire category of legal risk. Deploying Matomo for public sector clients and regulated industries is often done specifically because no data leaves the infrastructure. The compliance simplification is worth the operational overhead for these specific use cases.

Feature breadth is impressive for a self-hosted platform: heatmaps, session recordings, funnel analysis, A/B testing, form analytics, media analytics, cohort analysis. This is functionality that typically requires 3-4 separate SaaS subscriptions. The quality varies — heatmaps are functional but not as sophisticated as Hotjar's, A/B testing lacks the statistical rigour of dedicated experimentation platforms — but having everything in one data-sovereign installation is operationally convenient.

Performance at scale is Matomo's recurring challenge. MySQL struggles above 10 million events per month without significant optimisation: archive cron configuration, custom indexing, table partitioning, and eventually a move to ClickHouse. Large Matomo installations can require weekly DBA intervention to keep response times acceptable. The ClickHouse migration path improves this dramatically but requires additional infrastructure expertise.

The API is comprehensive but reflects PHP-era design patterns. It's functional for automated reporting and data extraction, but integrating Matomo data into a modern data stack (BigQuery, dbt, Looker) requires custom ETL work. There's no native BigQuery connector, no streaming export, and no webhook-based event forwarding. You're building that integration yourself or using third-party connectors of varying quality.

The plugin ecosystem is Matomo's double-edged sword. Community plugins extend functionality significantly, but quality control is inconsistent. Unvetted plugins can degrade performance, break upgrades, or introduce security vulnerabilities. Vet plugins thoroughly, test in staging, and maintain a conservative plugin policy for production installations.