Shipping and specialties
Launch tiers
Match launch ceremony to blast radius, not to how hard the code was.
Not every ship deserves a blog post, a sales deck, and a war room. A launch tier is a shared word for how much coordination a change needs, so marketing, support, engineering and PM agree on the size of the event before work starts.
Concept
A launch tier is a contract about coordination cost, not about engineering difficulty. A hard infra migration can be quiet externally. A one-week UI change can be company-visible if it redefines the product story.
The big idea
Pick the tier in the PRD or kickoff, not the week before ship. The tier decides whose calendar you need and which checklist rows are mandatory.
| Tier | What it is | Typical effort | Examples |
|---|---|---|---|
| T0 / dark | Internal or staff-only; no external promise | Flag, monitoring, rollback | Dogfood a redesign; shadow a ranking model |
| T3 / quiet | Small improvement for existing users | Release notes, in-app tip, support macro | New table filter; copy fix; minor API field |
| T2 / feature | Meaningful capability for a known segment | Beta, docs, sales one-pager, support training | Self-serve export; new integration; pricing add-on |
| T1 / major | Company-visible bet or new surface | Full GTM: campaign, press, exec review, multi-week beta | New product line; core workflow redesign; PLG free tier |
Over-tiering burns people on ceremony. Under-tiering ships into a vacuum.
Trade-off
Default to one tier lower than your excitement suggests, then raise it if sales, support or brand risk is real. Ceremony without risk is vanity. Risk without ceremony is how launches fail in public.
Walking through a choice
You spend six weeks migrating search onto a new index. Users see the same box. Errors and latency are the only things that can go wrong. That is T3 or even T0: feature flag, dashboards, rollback, a changelog line. No launch event.
You spend one week adding a free tier that changes who the product is for. Sales, billing, support macros, a blog post and a beta all fire. That is T1, even though the diff was small.
Common mistakes
- Tiering by pride in the diff. Hard work can still be a quiet launch.
- Picking the tier the week before ship. Calendars are already full.
- Treating T1 as the default. Most changes are T3.
Check yourself
A six-week infra migration that users never notice. Which tier? T0 or T3. Coordination cost follows external blast radius, not engineering difficulty.
You want a blog post and a sales one-pager for a new export button. What should you do first? Ask whether sales, support or brand risk is real. If not, drop a tier.
Key takeaways
- A launch tier is a contract about coordination, not about how hard the code was.
- Pick the tier at kickoff so the right people and checklist rows are in the plan.
- Default one tier lower than excitement suggests, then raise it for real risk.
Further learning
Next, learn before you announce, in alpha and beta. Then see the map in launching.