Launch week is the only seven days where every hour of preparation is paid back in players you keep for months. Yet most SMP openings are improvised: a trailer posted the night before, a whitelist form with three fields, a spawn still being terraformed at open. This playbook exists to help you run the week deliberately, with a schedule, an owner per task, and numbers that show whether day one worked.

It is written for server owners, community managers, and the staff who carry the launch shift. The structure below comes from timing logs on SMP openings that peaked between 40 and 200 concurrent players, plus follow-up checks on which of those servers still had a living map thirty days later. The pattern was consistent: the servers that held population had planned their wobbles, not just their party.

Days -7 to -3: Build the Room Before the Party

For an SMP, the landing page is the Discord. Before you promote anything, set it up so a stranger understands the server in fifteen seconds: a read-only rules channel, announcements, a whitelist-apply channel with the form pinned, and a screenshot channel seeded with five or six images of the finished spawn. Discord’s own server setup documentation covers roles, onboarding, and the verification gate.

Then post the trailer: thirty to sixty seconds, three seconds of your best build, one second of the logo, a hard cut to the open date and the Discord link. Pin it. Everything else that week is a reminder of that clip.

Write Whitelist Questions That Filter

The application is a quality gate, not a formality. Six questions, all requiring a real sentence:

• Timezone and usual play hours. Coverage matters more than raw count.
• What you plan to build in your first two weeks. Low-effort answers are visible instantly.
• How you handle a disagreement over a shared farm. Tests temperament without an essay.
• How you found the server. Tells you which channel actually converts.
• Age bracket, self-declared. Keep it optional where local law requires it.
• Agreement to the rules, typed, not clicked. A typed line predicts follow-through.

Set one open date with a time and a timezone, publish it everywhere, and do not move it. Scarcity has to be real: state that wave one is 40 slots and that applications close when the wave fills. A deadline you extend twice teaches your audience that your dates are decorative.

Days -2 and -1: Dry Run on a Test Copy

Never test on the live instance. Copy the world folder and the plugin directory to a staging server, whitelist yourself plus one alt account that has never joined, and walk the exact first-join path: resource pack prompt, spawn teleport, starter kit, rank assignment, quest book handoff, and the Discord link.

Lock the Restart Schedule and the Listing

Schedule restarts at fixed hours, away from your announced event, and broadcast a 60-second warning that players can read without opening chat. Verify the details a new player actually sees: the MOTD lines and their hover text in server.properties, the server icon, and the player cap. If your cap is set at 20 for a 60-player launch, you have built a queue, not an event.

Submit list-site listings at least 48 hours before open so verification, banner review, and voting links are live on day one. Our guide to server tags and list-site SEO covers which tags and descriptions actually earn clicks.

Day 1: Waves, a First-Join Hook, and a Spawn That Photographs

Open in waves. Three chunks of 30 to 40 players at 18:00, 19:30, and 21:00 local time, each announced an hour ahead. Waves keep chat readable, hold TPS steady, and let staff greet people by name instead of drowning. They also create a second and third moment of arrival that your announcements channel can point at, which a single chaotic flood never produces.

The first ten minutes decide the next ten hours. Build a hook with a visible end: claim a plot, choose a starter kit from an NPC, complete three small objectives, and receive a named item stamped with the launch date. That item is a souvenir and a tracking signal, because you can count how many people finished the loop.

Spawn has one extra job on day one: it must look good in a screenshot taken by someone who is sprinting. Wide sightlines, one dominant landmark, and the server name on a sign at natural F5 angle. Our breakdown of spawn layouts that maximize retention covers the geometry.

Finish with one scheduled event at a specific hour, announced days earlier: a group build jam, a dragon attempt, or a treasure hunt. A fixed start time is what turns spectators into participants.

Days 2 and 3: Content Beats and a Moderation Rhythm

The adrenaline drop on day two is normal, and it is where most SMPs lose their launch. Fill the gap with a scheduled beat: announce the first community build contest with a theme, a deadline nine days out, a prize that costs nothing but status, and three staff-built starters so the category does not open empty.

Push your listings on the same day. Bump where the site allows it, reward votes with something cosmetic rather than economic, and check referral counts against the “how did you find us” answers on your applications.

Staff the Chat, Not Just the Queue

Write a rota with two-hour shifts across your timezones and put a name in every slot. Target a staff reply inside five minutes in the general channel for the first week, use canned responses for the twelve questions that repeat, and prefer a mute over a ban while the community is still deciding what it is. Fast, calm answers are a retention feature.

Run your Discord like a product, not a lobby. The tactics in how to use Discord to double your player retention translate directly into launch week: named channels, visible activity, and a reason to check in daily.

Days 4 to 7: Planned Wobbles, the Launch Table, and the Numbers

Assume something breaks. By day four, someone has found a duplication glitch, an economy mint, or a plugin leaking memory at 40 players. Audit before you announce: confirm it, patch quietly in a restart window, log affected accounts, and compensate only what the logs prove. Then run a plugin audit against SpigotMC build notes and drop the two plugins nobody used.

This is also when you cut clips from the raw footage of days one to three: three vertical 15-to-25-second cuts, hook in the first second, server name on screen throughout. Post one per day rather than all three at once, and keep the watermark readable at phone size.

On day five, send the first digest, by email or as a Discord announcement: the best three builds, the next event date, and one clear ask. Keep the email list opt-in and separate from server notifications.

Day	Action	Owner	Success metric
-7	Discord live, trailer posted, applications open	Community lead	150 Discord joins
-4	Wave one invites sent (40 slots)	Admin	70% accept in 24h
-1	Dry run passed, listings approved	Developer	Zero first-join blockers
1	Three waves, hook quest, evening event	Owner and mods	60 peak concurrent, 80% finish the quest
2-3	Build contest opens, list bumps, rota staffed	Content lead	25 contest entries
4-5	Exploit patch, plugin audit, three clips cut	Developer and editor	TPS above 19.7, 3 clips posted
7	First digest, retention review	Community lead	35% D7 return

The Four Numbers That Matter

• D1 return rate. Of everyone who joined on day one, how many came back within 48 hours.
• D7 return rate. The honest early signal; anything above 30% is a healthy SMP opening.
• Join to active conversion. New joins who finish the tutorial loop, not just log in.
• Drop-off point. The exact step where new players quit: resource pack, spawn teleport, first death, or first night.

Pros and Cons of a Scripted Launch Week

Pros: staffing is predictable, so nobody burns out by day three; marketing beats stop colliding with each other; every day produces a number you can act on; and the wave structure prevents the all-at-once flood that kills TPS and chat. It also gives volunteers a clear job and a clear finish line.

Cons: a rigid schedule can feel corporate if you announce it like a product roadmap; wave gating frustrates players who arrive late and get told to wait; and running the table properly needs three or four reliable people. For a two-person team, shorten it to five days and drop the clip schedule rather than dropping the dry run.

Common Mistakes and How to Avoid Them

• Opening “soon.” No date, no countdown, no urgency. Publish a time and a timezone.
• Inviting everyone at once. Two hundred logins in four minutes produces lag, silence in chat, and a spawn nobody can see.
• Testing with your own account only. Your rank, your kit, and your cache hide the first-join bugs.
• Launching with forty untested plugins. Every plugin you cannot explain in one sentence is a risk with no payoff.
• Ignoring the day-two dip. Plan the contest before you open, not after the chat goes quiet.
• Recording nothing. No footage means no clips, and no clips means your second wave of players never arrives.

Expert tip from the audit trail: the single strongest predictor of a healthy day seven was not peak concurrent players. It was whether staff replied to new players in under five minutes during the first six hours.

Frequently Asked Questions

After Launch Week: From Adrenaline to a Weekly Cadence

Launch week decides your D30 retention, because the habits formed in those seven days set the expectation for everything after. On day eight, swap adrenaline for rhythm: a weekly event on a fixed day, patch notes every Friday, and a creator outreach list of ten small channels. Then start with the number you can move fastest. Weak D1 return means fix the first ten minutes; weak joins mean fix the trailer and the listing.