// the take
Dub is the open-source core behind dub.co, a link shortener with real attribution tracking (clicks, leads, sales) rather than just click counts. It's aimed at teams that want Bitly-style short links but need to tie clicks through to actual conversions and revenue, and are willing to self-host a fairly involved stack to get there.
Analytics run through Tinybird (ClickHouse under the hood) instead of counting rows in Postgres, so click/event volume that would choke a naive schema actually scales. The attribution chain is end-to-end — click, lead, and sale tracking routes are all wired to Stripe/Shopify/HubSpot webhooks, which is rare for an open-source link tool and is usually the hard part teams build themselves. It's a real production app (250M+ clicks/month per the README), not a toy, and the commit history backs that up with daily activity. Multi-domain custom link routing via the [domain]/[key] app-router structure is a clean way to handle white-labeled short domains.
Self-hosting means running PlanetScale (or MySQL-compatible equivalent), Upstash Redis, and Tinybird alongside Next.js — there's no single-binary or docker-compose-and-done path, so the operational cost of self-hosting is real. The AGPLv3/EE split puts SAML SSO, SCIM provisioning, and domain registration behind a commercial license, so the parts enterprises actually ask for aren't in the open core. Vendor coupling is heavy: swapping out Tinybird for analytics or Stripe for billing would mean touching a lot of webhook and cron code, not just a config flag. With this much surface area — dozens of cron jobs, multiple payment/ecommerce webhook integrations, domain renewal flows — the blast radius of a self-hosted instance misconfiguring billing or domain transfer logic is high, and the README doesn't point to test coverage beyond a Playwright e2e setup.
We read the README, the directory tree and a key source file in October 2026. Nobody pays to be listed. How we judge →