Wipe 101

The roster boss: the only encounter in TBC with no loot table

Why you keep coming up two short, the attendance maths nobody does, and how to use pugs, recruitment and retention in the order that actually works.

In brief. You probably do not have a recruitment problem. You have an attendance problem wearing a recruitment problem's clothes, and adding people to it makes the bench worse without fixing anything. Do the maths first: to field 25 you need 30 to 32 on the roster. Then pugs for tonight, recruitment for next month, and retention for the only fix that keeps paying.

I have cancelled a raid at 19:41 with twenty-three people online. Twenty-three. Two of them were in Shattrath "just grabbing flasks", one was on an alt insisting he would be two minutes, and the twenty-fifth person had posted "sorry, kid's sick" in Discord at 19:38.

We did not wipe that night. We just did not raid. That is the roster boss, and it has killed more TBC guilds than Vashj ever did.

What is the roster boss?

It is the point where your guild stops raiding because it cannot field 25 people, not because it cannot kill anything. No enrage timer, no strategy video, no loot. Just a Tuesday where the raid quietly does not happen, and then another one.

The cruel part is that it does not feel like a crisis. A wipe feels like a problem you can solve — you go read something, you change the assignment, you try again. Twenty-three people online feels like bad luck. Six weeks of bad luck later, your best healer has quietly joined the guild that raids on Wednesdays.

Why is 25-man so much worse for this than 10-man?

Because 25 is a hard floor, the roles are not interchangeable, and TBC gates substitution behind attunements. A 10-man can run at nine and shrug. A 25-man at twenty-two is a different raid.

The three things stacking up:

  • The number is inflexible. Drop below 25 and you are not just missing damage. You are missing a battle rez, a Fear Ward on the wrong tank, or the second Shaman whose Bloodlust you built the burn phase around.
  • Roles do not substitute. You need about two tanks, five to seven healers, and the rest DPS. Healers take the longest to recruit and the longest to gear. Nobody has ever posted "LF guild, geared healer, very flexible on times" and gone unanswered for more than four minutes, which tells you everything about the market you are shopping in.
  • Attunements block the easy patch. Your mate who wants to help cannot just come to SSC. He needs the questline, and if you are in Hyjal he needs Vashj and Kael dead. The barrier that made progression feel earned is the same barrier that stops you filling a gap on a Tuesday.

There is a fourth one nobody admits: the raid ID. Half your emergency pug list is already saved to the lockout you need them for, because they pugged it Sunday. Ask before you promise the raid you have someone.

How many raiders do I actually need?

Thirty to thirty-two. Not twenty-five, not "as many as I can get" — thirty to thirty-two, and the reason is that raid attendance is a percentage, not a headcount.

Here is roster size times realistic attendance:

Roster size85% attendance90%95%
2521.322.523.8
2823.825.226.6
3025.527.028.5
3227.228.830.4
3529.831.533.3

Look at 30 raiders at 85%. That is an average of 25.5, which reads like "just enough" and is in fact a disaster.

An average of 25 means you are short in roughly half of all weeks. That is what an average is. Half your weeks land underneath it. Half-term, a cup final, three people who all have the same cold because their children all go to the same school — none of that averages out on the night. It just cancels the raid.

So you want your average comfortably above 25, which means:

  • Under 28 on the roster: you are one virus away from calling it.
  • 30 to 32: an average of 27 to 29. You can lose four people to a bad week and still pull.
  • Over 35: now you have a different problem. At 90% attendance you are telling six people every week that they are not raiding, and benched raiders do not stay benched. They leave.

That last line is the one over-correcting guilds miss. Recruiting harder is not free. Every raider past about 32 is a person you have signed up to disappoint on a regular schedule.

Do I actually need to recruit, or do I have a different problem?

Probably a different problem. Recruiting is the reflex and it is usually wrong, so match the symptom before you write the post.

What you seeWhat it usually isWhat actually fixes it
Short two or three most weeksRoster genuinely too thinRecruit
Short only on one nightA schedule clash, not headcountMove the night
Short on healers, never on DPSRole imbalanceTargeted recruiting, and offer to gear them
28 sign up, 22 turn upAttendance and engagementFix the raid, not the roster
Raids are full, but new people vanish after a monthRetention and onboardingFix the bench policy and the trial

If you are in one of the bottom two rows, recruiting will make it worse. You will add people to a raid that is not worth turning up to, they will also stop turning up, and you will conclude that you need to recruit harder. I have watched a guild do three consecutive recruitment pushes without once asking why the people already there had stopped coming.

Should I just pug it?

Tonight, yes. As a plan, no. A pug fixes this Tuesday and nothing else, and a guild pugging six spots every week has quietly stopped being a guild — it is a raid ID with a tabard.

Used deliberately though, pugs are the fastest tool you have, and a good pug list is worth building on purpose rather than in a panic at 19:35.

How do I vet a pug in two minutes?

Look for evidence of preparation, not big numbers. Preparation predicts behaviour. Damage predicts one good night.

What I actually check, in order:

  • Consistency, not peaks. One 95th percentile parse sitting next to five in the twenties tells me about one fight where the boss stood still, not about a raider.
  • Enchants and gems on their last raid. This is the highest-signal thing available. Someone who turns up fully enchanted for a pug will do it for you. Three empty sockets and a green weapon enchant means you will be managing them.
  • Whether the meta gem is actually switched on. This is my favourite tell, because it is invisible in-game unless you look. TBC meta gems need a colour requirement met — so many reds, so many blues — and if you miss it the gem simply does nothing. Half the raid is walking around with an inactive Relentless Earthstorm Diamond and no idea. It is not a competence signal exactly. It is a "does anyone here check anything" signal.
  • Are they saved? Ask before the invite. "Are you saved to SSC this week" has retrieved more evenings than any other question I know.
  • Have they killed it? An honest "no, but I have watched it twice" is worth more than a confident yes from someone who has seen phase one.

This is the one job the log auditor does that has nothing to do with your own raid: paste an applicant's last report and the preparation questions answer themselves in about thirty seconds. Enchants, gems, whether they flasked, whether the meta was live.

How do I make pugs want to come back?

Treat them like a raider and tell them when you raid next. The best pug is one you have used before, and a good list builds itself if you are not awful.

  • Give them loot on the same rules as everyone else, and say the rules out loud before the first pull, not when the item drops.
  • Put them in a real spot. Do not park the pug in the group nobody wanted.
  • Add them on Discord rather than just inviting and forgetting.
  • Tell them your raid nights. Most pugs are guildless raiders looking for approximately what you already have.

What actually makes people apply?

Specifics, including the unflattering ones. Every recruitment post says "semi-hardcore, friendly, need healers" and describes four hundred identical guilds. Nothing in it is checkable, so nothing in it is persuasive.

The things worth writing down:

  • Exact times, including the end time. "19:30 to 22:30 server, invites at 19:15, we stop at 22:30 even mid-pull" beats "two nights a week" by a mile. The end time is the bit people are actually asking about.
  • Your real progress, including the boss you are stuck on. Say Vashj. The people who want to help you kill Vashj cannot find you if you write "SSC progression".
  • Your loot system in one sentence, and who settles arguments.
  • What you expect. Flasks, enchants, having watched the video. Concrete things.
  • What you do not expect. "We do not run farm nights" and "there is no attendance spreadsheet" are genuine selling points to exactly the person you want.
  • How many you need and in what role. "Two healers, and one of them is a raid spot rather than a bench spot" converts. "Always recruiting exceptional players" does not, and never has.

How do people find me when they are not reading forums?

They search, and they filter. Most recruitment happens when someone goes looking on a Sunday afternoon, not when they happen to read your post the day you make it.

That means being where people look: your realm's Discord, the forums, and directories that list guilds by what their reports show rather than by what they wrote about themselves. Your entry in the guild directory builds itself from your own logs, so it does not go stale when you forget to update it — and if you are hiring, flag yourself as recruiting, because that is a filter people actually use.

Why do people keep leaving?

Almost never because of a wipe. They leave because of the accumulated small signals that their evening was not respected, and every single one of them is fixable.

The list, roughly in order of how often I have seen it:

  • The raid starting twenty minutes late, every week, with no comment.
  • Wiping to the same mechanic in week four with nothing about the plan changed.
  • Sitting on the bench with no rotation, no warning and no explanation.
  • Loot going to whoever argued hardest.
  • Nobody noticing they had been running a broken spec, or an inactive meta gem, for a month.

That last one deserves its own paragraph. People leave guilds where getting better is invisible. On a progression fight the only feedback available is "did the boss die", and for several weeks the answer is no. If that is the only signal in the building, then as far as anyone can tell, nothing they do matters.

The fix is not a spreadsheet and a lecture. It is being able to say "that pull was genuinely cleaner, here is the bit that changed" and be right about it. That is what reading your own logs is actually for — not naming and shaming, which nobody has ever enjoyed, but having something true to point at when people need to know the evening was worth it.

What do I do this month?

In this order, one thing per week. Doing them out of order is how the recruitment push gets wasted.

  1. Week one — count. Sign-ups against actual arrivals, last four raids. You now know whether you have a roster problem or an attendance problem. Do not skip this. Everything below branches on the answer.
  2. Week two — patch and plug. Build a list of five pugs you would use again. In parallel, fix the cheapest attendance leak you found — it is almost always the start time, the end time, or the bench policy.
  3. Week three — recruit, narrowly. Post for the exact roles the maths demands, with real times and honest progress. Aim at 30 to 32, not "everyone who applies".
  4. Week four — make it worth attending. Audit a full clear and fix the free stuff: enchants, gems, consumables, the meta gems nobody switched on. It costs gold rather than skill, and a raid that visibly got better is the best recruitment post you will ever write.

The short version

  • 30 to 32 raiders for a 25-man. Under 28 you are fragile; over 35 you are benching people until they leave.
  • An average of exactly 25 means cancelling half the time. Build headroom, not parity.
  • Diagnose before recruiting. If 28 sign up and 22 arrive, more people will not help.
  • Pugs buy tonight. Vet on preparation, ask whether they are saved, then be nice enough to reuse them.
  • Recruitment works when it is specific, especially about the end time.
  • Retention is the only lever that compounds — and the fastest way to lose someone is to make improvement invisible.