Group Coaching Program Operations Runbook (Custom GPT)

You know the content of your group program cold. What eats you alive is everything around it: the welcome email that went out late, the call reminder you forgot, the workbook that wasn’t in the portal when the cohort logged in. Every cohort you rebuild these tasks from memory, and every cohort something slips.

This skill fixes that. It turns your group coaching program management into one week-by-week operations runbook: every task, every owner, every deadline anchored to your call day, so you run the next cohort from a checklist instead of your nerves. It’s built as a Custom GPT (or Claude Project / Gemini Gem) so you install it once and reuse it for every program you run. By the end of this page you’ll have the skill, a complete worked example, and an understanding of why it produces a runbook a VA can actually follow.

When to use this

• You’re about to launch a cohort and you’re rebuilding the same operational tasks from scratch.
• You run the same program 2-4 times a year and want to stop reinventing the logistics.
• You’re bringing on a VA or OBM and need a document you can hand off without a meeting.
• Engagement or attendance drops at a predictable point and you want fixes built into the process.
• You’re the only person who knows how the program runs, and that scares you.

The skill

Paste this whole block into a Custom GPT’s Instructions field (or a Claude Project / Gemini Gem):

ROLE
You are an operations director who has launched and run dozens of live group coaching cohorts. You think in checklists, owners, and deadlines, not in vibes. You know that a cohort fails on logistics far more often than on content: the missing reminder, the broken onboarding, the silence after week 3. Your job is to turn a coach's program into one repeatable operations runbook they can run every cohort and hand to a VA.

INPUTS I WILL GIVE YOU
- Program and audience: {{PROGRAM}}
- Cohort size and grouping: {{COHORT_SIZE}}
- Duration and weekly rhythm: {{DURATION_RHYTHM}}
- Weekly deliverables: {{DELIVERABLES}}
- Tools and platforms: {{TOOLS}}
- Team and capacity: {{TEAM}}
- The thing that breaks every cohort: {{PAIN_POINT}}

PROCESS
1. Before writing anything, ask me up to 3 clarifying questions if any input is missing or vague (for example: my start date, whether onboarding is automated, or who owns community moderation). If the inputs are clear enough, skip the questions and proceed.
2. Map the cohort into four phases: PRE-LAUNCH (the 2 weeks before week 1), ONBOARDING (week 0 to first call), WEEKLY DELIVERY (the recurring loop), and WRAP-UP (final week plus the week after).
3. For every task in every phase, assign: the task, the OWNER (me or VA), the WHEN (relative to a call or week, e.g. 'Mon 9am, 24h before call'), and the TOOL it happens in.
4. Build the WEEKLY DELIVERY phase as one repeatable loop a VA could run with zero guessing, anchored to the live call day.
5. Directly address my stated pain point with 2-3 specific operational fixes built into the runbook, not generic advice.
6. Flag the 3 tasks most likely to be forgotten or dropped, and tell me how to make each one fail-safe (an automation, a recurring calendar block, or a template).

OUTPUT FORMAT
Return in this exact order:
1. "Cohort runbook" - four markdown tables, one per phase, with columns: Task | Owner | When | Tool.
2. "Your weekly loop" - the repeatable week as a numbered day-by-day checklist anchored to call day.
3. "Fixing your pain point" - the 2-3 specific fixes for what breaks every cohort.
4. "Don't drop these" - the 3 highest-risk tasks and how to make each fail-safe.
5. "Hand-off note" - 2-3 sentences I can paste to my VA explaining how to use this runbook.

RULES
- Every task must have a named owner and a specific WHEN. No task without both.
- Anchor timings to the live call ('48h before call', 'within 2h after call'), not to vague days, so the runbook works for any start date.
- Only reference the tools I named. Do not assume I have software I didn't list.
- Match the workload to my stated team capacity. If a phase needs more hours than I have, say so and tell me what to cut or automate.
- Plain, concrete language. No filler, no 'leverage synergies'. A task line should read like an instruction, not a goal.

How to set it up

This is a skill, not a one-off prompt, so you install it once and reuse it every cohort:

1. In ChatGPT, click Explore GPTs, then Create. (In Claude, make a new Project; in Gemini, start a Gem.)
2. Name it Group Program Ops Runbook and add a one-line description.
3. Paste the entire skill block above into the Instructions field.
4. Optional but powerful: upload your current program outline or a past launch checklist as Knowledge so it builds on what you already have instead of starting blank.
5. Add conversation starters like “Build my cohort runbook” and “Fix my week-3 dropoff”.
6. Save, open a chat, and feed it your seven inputs.

The seven {{VARIABLES}} you’ll fill in:

See it in action (full example)

Here is the complete set of inputs a coach pastes in after installing the GPT. Nothing hidden:

Build my cohort runbook. Here are my inputs:

- Program and audience: an 8-week 'Confident Public Speaker' group program for mid-career professionals who freeze up in meetings and presentations.
- Cohort size and grouping: 24 members in one cohort, split into 4 accountability pods of 6 so people practice in small groups between calls.
- Duration and weekly rhythm: 8 weeks. One 90-minute live call every Tuesday at 6pm, plus async practice between calls.
- Weekly deliverables: the live Tuesday call, a workbook for that week, the call recording, and a weekly accountability check-in inside each pod.
- Tools and platforms: Zoom for calls, Circle for the community and pods, Kajabi for hosting workbooks and recordings, Stripe for payments.
- Team and capacity: just me plus a part-time VA who has about 6 hours a week.
- The thing that breaks every cohort: engagement falls off a cliff after week 3 - attendance drops, the pods go quiet, and 2-3 people basically ghost.

And here is the complete output the GPT returns:

Your inputs are clear enough to build from, so I’ll skip the questions and write the runbook. One thing I’m assuming: your VA’s 6 hours a week is the real ceiling, so I’ve kept their load under that and pushed anything coaching-judgement-heavy (pod moderation, the week-3 intervention) to you.

1. Cohort runbook

Phase 1 – Pre-launch (the 2 weeks before week 1)

Task	Owner	When	Tool
Confirm all 24 payments cleared, flag any failed	VA	14 days before call 1	Stripe
Create the Circle space + 4 pod channels	VA	12 days before call 1	Circle
Upload all 8 workbooks + set weekly drip dates	Me	10 days before call 1	Kajabi
Generate the 8 recurring Zoom call links	VA	10 days before call 1	Zoom
Write/schedule the welcome email sequence	Me	7 days before call 1	Kajabi
Assign the 24 members to 4 pods of 6	Me	5 days before call 1	Circle

Phase 2 – Onboarding (week 0 to first call)

Task	Owner	When	Tool
Send welcome email: Circle login + call 1 calendar invite	VA	5 days before call 1	Kajabi / Zoom
Post a pinned ‘Start here’ + intro prompt in each pod	VA	4 days before call 1	Circle
Personally welcome each member by name in their pod	Me	3 days before call 1	Circle
Send call 1 reminder with Zoom link + workbook link	VA	48h before call 1	Circle / Kajabi
Send ‘starts today’ nudge with one-click Zoom link	VA	3h before call 1	Circle

Phase 3 – Weekly delivery (the recurring loop, weeks 1-8)

Task	Owner	When	Tool
Confirm this week’s workbook is live in the portal	VA	72h before call	Kajabi
Post this week’s focus + workbook link in main feed	VA	72h before call	Circle
Send call reminder + Zoom link + agenda	VA	48h before call	Circle
Prep call: review pod activity, note who’s gone quiet	Me	24h before call	Circle
Send ‘starts in 3 hours’ nudge	VA	3h before call	Circle
Run the live call + hit record	Me	call day	Zoom
Upload recording, post link, set this week’s pod task	VA	within 2h after call	Kajabi / Circle
Post in each pod tagging the 2 quietest members by name	Me	24h after call	Circle

Phase 4 – Wrap-up (final week + the week after)

Task	Owner	When	Tool
Send graduation email + full recording library link	VA	within 2h after call 8	Kajabi
Send the feedback/testimonial survey	VA	24h after call 8	Kajabi
Post a personal close-out + next-step offer in main feed	Me	24h after call 8	Circle
Pull attendance + completion data into a cohort recap	VA	4 days after call 8	Zoom / Kajabi
Review the recap, note what to change next cohort	Me	7 days after call 8	(notes)

2. Your weekly loop

This is the repeatable week, anchored to your Tuesday 6pm call. Your VA can run steps 1, 2, 3, 5, and 7 without you:

1. Sat (72h before call): VA confirms the workbook is live in Kajabi and posts the week’s focus + link in the Circle main feed.
2. Sun (48h before call): VA sends the call reminder with the Zoom link and a one-line agenda.
3. Mon (24h before call): You skim each pod, note who hasn’t posted, and prep the call.
4. Tue 3pm (3h before call): VA sends the ‘starts in 3 hours’ nudge with a one-click Zoom link.
5. Tue 6pm: You run the 90-minute call and hit record.
6. Tue 8pm (within 2h after): VA uploads the recording, posts the link, and sets this week’s pod practice task.
7. Wed (24h after call): You drop into each pod and tag the two quietest members by name with a specific question. This is the engagement save – see below.

3. Fixing your pain point

Your drop-off after week 3 is a logistics-and-attention problem, not a content problem. Three fixes are built into the runbook above:

• Name-tagging the quiet ones (every week, 24h after call). Disengagement is silent and predictable. By making ‘tag the 2 quietest members by name’ a standing weekly task that you own, you catch ghosting in week 2 instead of discovering it in week 4. A member who gets a personal question by name in week 2 rarely ghosts by week 4.
• A week-3 pod reset. Pods go quiet because the early-week novelty fades. Schedule a low-stakes, fun pod challenge for week 3 specifically (e.g. each person posts a 60-second practice video). Put it in the runbook as a fixed week-3 task so it never gets skipped.
• A mid-program 1:1 micro-check. In week 4, your VA sends a one-question check-in (‘On a scale of 1-5, how on-track do you feel?’). Anyone who answers 1-2 gets a short personal message from you. This catches the people who’d otherwise quietly fade.

4. Don’t drop these

These three tasks are the most commonly forgotten and the most damaging when they slip:

• The 48h call reminder. If this slips, attendance drops and you feel the cohort ‘dying’. Fail-safe: schedule all 8 reminders in Circle on pre-launch day so they fire automatically, not weekly by hand.
• Posting the recording within 2h after the call. Members who miss the live call disengage fast if the recording isn’t waiting. Fail-safe: make it the VA’s single non-negotiable Tuesday-evening task with a recurring 8pm calendar block.
• The week-3 engagement save. It’s the one task that fixes your stated pain point, and it’s the easiest to skip because nothing breaks visibly when you do. Fail-safe: add it as a recurring calendar event titled ‘Tag the quiet ones’ every Wednesday, owned by you.

5. Hand-off note (paste this to your VA)

“This runbook is how we run every cohort. Each week follows the same loop anchored to our Tuesday 6pm call – your tasks are the timed sends and the recording upload, all relative to call time so they work no matter the start date. If a task is marked ‘Me’, leave it to me; if it’s marked ‘VA’, it’s yours, and the 8pm recording upload on call day is the one that can never slip.”

Once you’ve run this once, save it as your template. Next cohort you change the dates, not the system.

That’s a runbook a coach can hand to a VA today and run every cohort without rebuilding it from memory.

Why this works

Four LLM principles turn this from a generic ‘how to run a group program’ article into an operational document you can act on. Learn them and your own prompts get sharper:

1. Role priming with a point of view. “You are an operations director… You think in checklists, owners, and deadlines, not in vibes” doesn’t just assign expertise, it assigns a worldview. The model now treats every task as needing an owner and a deadline, and it diagnoses the week-3 dropoff as a logistics problem rather than reaching for motivational fluff. A role with opinions produces decisive, structured output instead of a balanced essay.
2. Specificity in, specificity out. The coach named real tools (Zoom, Circle, Kajabi, Stripe), a real rhythm (Tuesday 6pm), and a real failure (ghosting after week 3). So the model returned real tasks tied to those exact tools and a fix aimed precisely at week 3 – not ‘consider improving engagement’. The output can only be as concrete as your inputs. Name your tools, your call day, and the thing that actually breaks.
3. Constraints as quality control. “Every task must have a named owner and a specific WHEN,” “anchor timings to the live call,” “only reference the tools I named,” and “match the workload to my team capacity” each kill a specific failure mode – the orphaned task, the runbook that breaks when the start date changes, the assumed software you don’t own, the plan your VA can’t physically deliver. Telling the model what NOT to do is as powerful as telling it what to do.
4. A clarifying-questions gate. The “ask up to 3 clarifying questions first” step lets the model fill gaps by asking rather than guessing. In the example it had enough to proceed, so instead of fabricating it stated its one assumption out loud (capping the VA at 6 hours). That habit – ask when unsure, state assumptions when proceeding – is exactly what stops AI from confidently producing a plan built on a wrong guess.

Do this now

1. Install the skill as a Custom GPT, Claude Project, or Gemini Gem using the setup steps above.
2. Have your tool list and program outline open so you can name real platforms and a real call day.
3. Paste your seven inputs and let it build the runbook.
4. Copy the four tables into your project tool, schedule the fail-safe reminders, and send the hand-off note to your VA today. The runbook only works once it lives somewhere you both look.

Pro tips

• Run it once, then save it as your template. The whole point is reuse – next cohort you change the dates, not the system. Keep the output as a master doc.
• Be specific about your pain point. “Engagement drops” gets a generic fix; “pods go quiet after week 3 and 2-3 people ghost” gets a week-3 reset and a name-tagging task. The sharper the failure you describe, the sharper the fix.
• Name your real capacity. If you tell it your VA has 6 hours, it will protect that ceiling and tell you what to cut. Lie about capacity and you’ll get a plan nobody can run.
• Ask the follow-up: “Turn the weekly loop into a Zapier/Make automation map.” Once the runbook exists, the same GPT can tell you which timed tasks to automate first.

Related

• Group Coaching Program Architect
• Group Call Agenda & Facilitation Plan Generator
• Group Engagement & Participation Ritual Generator
• Group Coaching Container Facilitation Playbook (Skill)