App starts, login works, the top three journeys complete, integrations answer — 10–20 tests, under five minutes, ruthlessly maintained.
On merge (did we break the build), on deploy to staging (did config survive), and post-deploy to production (is it actually up) — the same cheap suite, three vantage points.
Focused verification around a specific fix or config change — smaller than regression, sharper than smoke, the right tool for the hotfix path.
Post-deploy verification with automatic alerting — the gap between 'deployed' and 'verified working' is where the worst incidents grow, silently.
Skipping the discipline this article describes until an incident, audit, or stalled project forces it — every practice above is cheaper adopted early than retrofitted under pressure.
Let's discuss how we can help you with smoke sanity release gates.