DataAI Nexus

Agents / Skills

Launch Checklist

Skill

Builds the checklist that makes a launch boring: everything that must be true before, during and after — across product, comms, support and measurement — with owners and go/no-go criteria, so launch day executes instead of improvises.

Best for

  • Launches that discover their gaps live ('did anyone tell support?')
  • Coordinating the product, marketing and support threads that each assume the others are handled
  • The go/no-go meeting that needs criteria instead of courage

What you give it

  • The launch: what ships, when, to whom, and the teams involved

What you get back

  • The checklist by phase: pre-launch (verified, not assumed), launch window (sequenced, owned), post-launch (the watch, the follow-through) — every item with an owner and a done-check
  • The cross-team layer most launches miss: support briefed with the FAQ, comms sequenced with the feature flag, measurement live before the traffic arrives
  • The go/no-go gate: the criteria that mean go, decided before the meeting where adrenaline votes

How it works

  1. Builds by phase with verification discipline: pre-launch items are verified done, not assumed done — 'the email is ready' and 'the email renders, links work, and is scheduled against the flag time' are different claims.
  2. Covers the cross-team seams where launches actually fail: support's brief and FAQ, comms' sequencing against the technical rollout, legal's sign-offs, measurement's readiness before traffic.
  3. Sequences the launch window explicitly: what happens in what order, owned by name, with the gaps between steps stated — the flag before the email, the smoke test before the announcement.
  4. Pre-decides the gate and the aftermath: go/no-go criteria written calm, the post-launch watch with owners for the first hours, and the week-after follow-through that launches forget.

Example

You: We launch the new plan structure Tuesday — product, pricing page, emails, the works. Build the checklist.

Result: The checklist found the gaps on Thursday instead of Tuesday: support had not been briefed and the refund-policy edge cases had no answers (FAQ written Friday, support walkthrough Monday), the announcement email was scheduled 9:00 but the feature flag was planned 10:00 (resequenced — the email that advertises what is not yet live is a classic), measurement for the new plan selection was not instrumented (added, verified with test events), and the rollback path for the pricing page had never been stated (one-line revert, documented, tested). Go/no-go criteria agreed Monday: the four greens required, listed. Tuesday: go called on criteria at 9:40, launch executed in sequence, the one surprise (a cache serving old prices for 20 minutes) caught by the post-launch watch and handled by its listed owner. Boring — as designed.

Limits — please read

  • A checklist encodes the knowable; the post-launch watch exists because surprises do not read checklists.
  • Scale it to the launch: the full apparatus for the pricing change, a one-pager for the minor feature — it sizes honestly.
  • The no-go must be sayable: criteria help, but the authority to call it is organisational.