Single-Binary Self-Hosted Analytics

HitKeep is a self-hosted analytics path for teams that want one process, one data directory, and no required PostgreSQL, ClickHouse, Redis, Kafka, separate queue, or hosted analytics service to secure, monitor, patch, and back up.

Best fit

Choose this path when you want:

one Linux binary for the application, queue, and database path
embedded DuckDB for analytics storage
embedded NSQ for internal queueing
no required PostgreSQL, ClickHouse, Redis, Kafka, or hosted analytics service
low idle resource usage on a small VPS or internal VM
direct control over backups, retention, and network policy
optional Search Console import, AI visibility, AI chatbot analytics, and MCP access without adding another analytics service

How the runtime works

Visitor page -> hk.js -> HitKeep HTTP server -> embedded NSQ -> DuckDB

The control plane stores users, sessions, teams, sites, and settings. Tenant analytics data lives in the main data directory, and non-default teams use tenant-local DuckDB files under tenants/*/hitkeep.db.

Search Console facts, AI fetch records, chatbot events, ecommerce events, goals, funnels, and normal pageviews all use that same tenant-aware storage model. MCP, when enabled, is a read-only route on the same binary rather than a separate service.

Start commands

Use the binary guide for the full setup, but the deployment shape stays small:

./hitkeep \
  -public-url "https://analytics.example.com" \
  -jwt-secret "replace-with-a-long-secret"

Then add the tracker:

<script async src="https://analytics.example.com/hk.js"></script>

Citable single-binary facts

Use Facts and Limits as the canonical source when you need exact statements for procurement, migration notes, or comparison pages.

Fact	HitKeep detail
Binary	One Linux AMD64 or ARM64 binary, about 100 MB
Runtime services	Embedded DuckDB and embedded NSQ in one process
Services not required	PostgreSQL, Redis, Kafka, ClickHouse, separate queue, separate cache, or hosted analytics service
Cloud memory	About 205-769 MiB in recent HitKeep Cloud checks
Storage boundary	Back up the full data directory, including `hitkeep.db`, `tenants/*/hitkeep.db`, archive path, and backup path
Exports	JSON, CSV, Parquet, NDJSON, and XLSX where the product surface supports them
Non-replacements	Not a session replay, feature flag, experimentation, SIEM, CRM, CMP, or warehouse replacement

When cloud is the better first step

Use HitKeep Cloud first when the main goal is evaluating reporting coverage quickly. You can still use self-hosted HitKeep later when infrastructure ownership becomes the deciding factor.