Rituals and calendar

A guild of 30 to 80 people is at or above the size where online groups stop feeling like friends and start feeling like a lobby. The one study that measured this across thousands of WoW guilds found the survivors were partitioned into small subgroups of about six, played together often, and ran scheduled activity rather than ad hoc questing. verified This page is our version of that: pods, a calendar of rituals, and a short list of numbers officers check once a month so decline is seen before it is felt.

Why this is built the way it is

Three findings drive everything below, all from the research file behind this page:

Small pods beat one big blob. Controlling for size, guilds whose largest interconnected subgroup was smaller survived better, and members advanced faster in guilds “fragmented into many cohesive subunits.” Optimal subgroup size was about six. verified
Belonging is the retention engine. Social motivation, not achievement, was the best predictor of hours played. People stay because they like their guildmates. verified
Attachment has to be manufactured socially, because the game gives you nothing. The healthy version of that is self-determination theory: people return to places where they feel competent, autonomous, and connected. Any ritual that makes people feel surveilled, shamed, or excluded reduces retention even if it spikes activity for a month. verified

The ethical filter for every mechanic is the Regret Test: if members knew everything the officers know about why a ritual exists, would they still take part? verified

Pods

Every member belongs to a pod of 5 to 7, formed by timezone and playtime, balanced by class where possible, and deliberately mixing friend groups so no pod is a pre-existing clique.
Each pod has a name (funny is fine here), a private thread in the wing, and a pod captain who is not an officer. The captain’s job is to say hello, notice who has gone quiet, and organize one thing a week. The reward is impact, and it is also the officer audition.
Pods reshuffle once a season (about every three months) so the guild stays one guild rather than five.
At launch, the leveling pods from the launch-plan thread are the first pods. Week 2 has a pod-merge check because half of launch pods stall. See the launch night playbook.
A pod going quiet is the earliest warning we have. Churn spreads through grouping networks: when your regular partners go quiet, you go quiet. verified

The calendar

Each row gives the ritual, who owns it, and in one line the psychology behind it. Owners are roles, not names; the Officer on duty rotates weekly.

Daily

Ritual	Owner	Why it works
Prompt of the day in the tavern (seven rotating prompts, dead ones retired monthly)	Bot	Lowers the bar to posting; lurkers reply to a question more than to silence
“Someone’s home” in the WoW Tavern voice channel, officers seed weeknights 8 to 9 pm	Everyone	Co-presence is the product: players value an audience even when not grouped verified

Weekly

Ritual	Owner	Why it works
Guild Night: one fixed evening, no raid, all welcome. Dungeons, PvP, leveling, or voice chaos	Officer on duty	The heartbeat. Raises both “play together” density and scheduled instance time, the two survival predictors we can move
Screenshot Friday thread: one image each, react only	Bot opens, members fill	Zero effort, high co-presence, no writing required
MVP of the Week: nominations Thursday, announced Sunday	Rotating non-officer picker	Recognition from peers, not from above; the rotation stops the same three people winning
Raid nights, maximum two for the core (when raids open)	Raid lead	Scheduled, predictable, roster 24 hours ahead. See Raid leading
Story of the Week: five minutes at the end of Guild Night, retold by whoever wiped worst	Wipe Historian	This is how lore forms. It also makes wiping a story instead of a failure
Weekly streak: “showed up to one guild thing this week”	Bot	See streak rules below

Biweekly

Ritual	Owner	Why it works
Newcomer coffee: 20 minutes in voice for anyone who joined in the last month, cancels itself if there is nobody new	Buddy coordinator	Newcomers’ top complaint is cliques from a previous game; this is the bridge across verified
Social event, alternating in-game and voice-only, with a guest host	Community officer plus guest host	Guest hosting is the cheapest engagement tool there is. See Contests and events
Farming night for whatever the crafter directory says we are short on	Bank Goblin	Counts as attendance; keeps the consumables bank alive without burning out crafters

Monthly

Ritual	Owner	Why it works
Guild Meeting in voice: 20 minutes maximum, agenda posted, decisions logged	GM	Guilds with codified procedures had happier members; this is also the leader’s heartbeat check verified
Hall of Deeds update and title awards	Officer	Public, named recognition. See Achievements and Ranks and titles
One contest, never two	Events volunteer	Rare enough to stay special; variable prize among all finishers, not the top one
Pod check-in: are all pods alive? Merge or split dead ones	Officers	Leading indicator of churn
Health metrics readout (table below)	Officers	Fifteen minutes. Decline seen before members feel it
State-of-the-guild post	GM	Closes the feedback loop publicly. See Officer handbook

Seasonal (quarterly)

Ritual	Owner	Why it works
Pod reshuffle, streak passes reset, season title awarded, season named	Officers	Prevents cliques hardening; gives the Chronicle chapters
Off-game night: board game site, another game from the parent server, a film	Volunteer	The longest-running guild on record credits life outside the game verified
Anonymous five-question survey, results posted	GM	People stay when satisfied; you cannot know that without asking

Yearly

Ritual	Owner	Why it works
Founding Day: in-game gathering, screenshot, awards, origin story retold	Everyone	Identity. The origin story is updated once a year, here
Meetup attempt, even a cameras-on call	Volunteer	One in-person meetup changed a top guild’s retention; opt-in only, public place, no pressure verified

Streak rules

Streaks work: the reference case is a language app whose 7-day-streak users were several times more likely to finish a course, and whose “streak freeze” cut churn among at-risk users. verified They also become guilt machines if built wrong. Ours:

The streak counts one guild thing per week: a guild night, a dungeon with guildmates, a screenshot post, or 30 minutes in voice. Never “raided every night.”
Two free passes a season, no explanation. The pass exists before the streak does.
Breaks are framed as progress: “47 of the last 50 weeks,” never “streak lost.”
No confirmshaming. Nobody gets a “are you really going to miss it?” ping. Ever.
Weekly, not daily. A daily streak in a game people play three nights a week is a trap.
The streak feeds the title layer only. It is never a public leaderboard.

Recognition rules

Leaderboards are bounded (one contest, one week, then archived), team-based (pod versus pod), effort-visible (dungeons run, mats donated), and opt-in. Never a global “messages sent” or “hours played” board: it rewards spam and shames the quiet. verified
A shoutout thread where anyone thanks anyone. Officers seed it with three real posts the first week; an empty channel tells newcomers the guild is dead.
Every new channel or thread is seeded with three real posts before it is announced.

Health metrics (monthly, 15 minutes)

Thresholds are proposals, not research findings. Adjust after two months of baseline. Red on any two rows triggers a short officer huddle, not a guild-wide announcement.

Metric	Source	Green	Yellow	Red
Active members (any in-game or Discord activity in 30 days) / roster	Bot + in-game roster	> 50%	35 to 50%	< 35%
Guild Night attendance (voice, 30+ min)	Bot	> 12	8 to 12	< 8
Pods with 3+ members active this month	Pod threads + voice	All	1 dead pod	2+ dead pods
New member 30-day retention	Bot	> 70%	50 to 70%	< 50%
Members who posted or spoke this month / active	Bot	> 60%	40 to 60%	< 40% (fine if stable)
Class balance (smallest class share of 60s)	In-game	No class < 5%	One class missing	Two classes missing
Level spread of mains	In-game	Wide	Narrowing	All-60, no alts leveling
Officer hours per week (self-report)	Officers	< 6	6 to 10	> 10
Drama tickets opened	Officer log	0 to 1	2	3+
Distinct MVP winners, rolling 12 weeks	Bot	10+	6 to 9	< 6
Returning Heroes this month	Bot	Any	0 for 2 months	0 for a season
Raid signups vs spots (when live)	Raid-Helper	110 to 130%	100 to 110% or > 150%	< 100% or bench > 30%
Survey “would you invite a friend”	Seasonal survey	> 8/10	6 to 8	< 6

Two rows deserve a note. Class balance was the single strongest survival predictor in the study: unbalanced guilds bleed their scarce class to guilds that are balanced. Level spread is a good sign, not a bad one: a guild that is all-60 with nobody leveling has stopped recruiting. verified

Unknown as of 2026-09-24

Where the first content drought lands depends on raid unlock timing, which is contested (December 9 versus some raids at launch). See Road to launch.
No academic study tests guild-level rituals directly. The streak and prompt evidence comes from consumer apps and Discord practitioners and is applied by analogy.
Whether Forever’s in-game roster exposes last-online data the way vanilla did, which the active-members row depends on. vanilla baseline