Operator context
Longer answer
The launch weekend runbook starts being drafted during discovery and gets refined through build and stabilisation prep. By cutover day, every team member knows their specific role, their checkpoints, and the conditions that would trigger a rollback. Brands that go into cutover without a fully documented runbook routinely have the runbook get written reactively during incidents.
The standard cadence: Friday evening — final pre-cutover backup, DNS preparation, source-platform freeze on new subscriptions. Saturday morning — DNS switch, redirect map activation, sitemap submission, first-hour validation. Saturday daytime — checkout validation, conversion monitoring, support team activation. Saturday-Sunday — ongoing monitoring, incident response, validation of subscription cutover at the first billing cycle event. Sunday afternoon — go/no-go on full traffic. The exact timing varies but the structure is similar across engagements.
Validation checkpoints punctuate the runbook. After each phase, named owner confirms the phase completed as expected before the next phase begins. The checkpoints are mechanical (verify DNS resolves correctly, verify redirects fire, verify checkout completes for a synthetic order) but mandatory; skipping them consistently lets small issues compound into customer-facing incidents.
Rollback triggers are pre-defined. "If checkout success rate drops below X% in the first hour, rollback DNS." "If 404 rate exceeds Y across high-traffic URLs, freeze redirects and investigate." Brands that have these triggers explicit before cutover make rollback decisions decisively when needed; brands without them debate the decision while customer impact accumulates.
Post-cutover, the first 72 hours have dedicated monitoring. Engineering on standby for technical issues. Customer service capacity elevated. SEO monitoring for 404 patterns. Marketing monitoring for analytics tracking. Each of these is a named owner with a defined response protocol. Brands that staff this window well absorb issues without customer impact; brands that staff it casually produce most of their post-launch incidents in these 72 hours.