Skip to content

FACILITATOR-GUIDE.md

Latest commit

 

History

History
121 lines (79 loc) · 5.77 KB

FACILITATOR-GUIDE.md

File metadata and controls

121 lines (79 loc) · 5.77 KB

Facilitator Guide

For anyone running a session from this curriculum — sellers, Solution Engineers, partner trainers, internal community leads, or learning facilitators.

The session format

Every session in this curriculum follows the same three-stage structure. Don't deviate without good reason — the structure is doing more work than it looks like.

Stage 1 — Pre-work (1–2 weeks before)

  • Send the participants the relevant Microsoft Learn module links from the track page.
  • Track these completions. If you can't see who completed pre-work, you can't recover the session in real time.
  • Send one scenario primer — a 1-paragraph description of the real workflow you'll work through live. Ask one prep question: "Where in this workflow do you currently spend the most time?"
  • For customer-facing sessions: confirm tooling access (Copilot license, Azure subscription, GitHub Copilot if relevant). Tooling-not-working in the first 10 minutes kills the session.

Stage 2 — Live session (60 minutes)

Minute Activity
0–10 Recap the pre-work concepts. Ask one question: "What surprised you?"
10–55 Hands-on with the scenario. Facilitator demos one example, then participants try a variant. Encourage live questions.
55–60 Q&A + the Responsible AI checkpoint (see below).

Responsible AI checkpoint — the last 5 minutes are non-negotiable. Ask three questions about whatever was just built:

  1. Privacy / fairness: What data did the AI see? Could it leak or bias?
  2. Human review: Where in this workflow does a human still validate the output?
  3. Failure mode: What's the worst that happens if this gives a wrong answer? Is the consequence proportionate?

This is the single most underused practice in AI training. Don't skip it.

Stage 3 — Post-work (the next 30–90 days)

  • Share the session recording (if recorded) and templates.
  • 30-day check-in: "Did anyone use what we covered?" Light touch — Teams message, email, or office-hours.
  • 90-day check-in: Measurable adoption — usage telemetry, time saved, workflow changed. This is your evidence for the next session pitch.

Session-level checklist

Before every session:

  • Pre-work sent ≥ 7 days in advance
  • Pre-work completion checked 2 days before (chase non-completers)
  • Scenario primer sent
  • Tooling access confirmed (for hands-on sessions)
  • Backup demo recorded (in case live demo fails)
  • Responsible AI checkpoint questions ready
  • 30-day and 90-day follow-up scheduled in your calendar now

Audience-specific guidance

Running for customers (sellers / SEs)

  • Lead with the customer's stated problem, not with the curriculum structure. "You said your team spends 3 hours a week on X. Today we'll cover the track that addresses that."
  • Don't pitch the product mid-session. This is enablement, not a demo. Trust builds faster when you focus on their learning.
  • End with a scoped follow-up offer: "I'd like to come back in 30 days and see what's changed. What metric should we look at?"
  • See adaptations/for-customer-workshops.md for the customer workshop format.

Running for partners

  • Train-the-trainer focus: the partner needs to be able to run this themselves after one session.
  • Provide the facilitator notes, slide source, and scenario library.
  • Cover why each step exists, not just what to do.
  • See adaptations/for-partner-enablement.md.

Running for internal teams

  • Use real internal scenarios. "How would we use Copilot for our actual quarterly planning doc?" beats any generic example.
  • Pair with an internal champion who'll be available between sessions.
  • See adaptations/for-internal-teams.md.

Three things that consistently go wrong

Across many cohorts, these are the recurring failure modes:

  1. No real scenarios. Generic examples produce generic engagement. Spend 30 minutes before the session surfacing one real workflow from the audience.
  2. Audience level mismatch. Half the room is at L100, half at L300. Run a 30-second self-assessment at the start ("raise your hand if you've already built a custom agent") and adjust on the fly. Or split the room.
  3. No follow-up. The session is forgotten in 14 days without a 30-day check-in. Schedule it before the live session, not after.

What to measure

For every session you run, track at least:

  • Completion rate — pre-work done / pre-work assigned
  • Live attendance — registered / attended
  • Activation — did anyone use the skill in the next 30 days? (1 question survey)
  • Adoption — did the workflow actually change in 90 days? (1 follow-up call or telemetry)

You don't need a dashboard. A spreadsheet with these four columns per session is enough to demonstrate impact and earn the next session.

When to deviate from the format

  • You have 30 minutes, not 60. Cut to 5 / 20 / 5 — recap, single hands-on, RAI checkpoint. Drop the Q&A; defer to follow-up.
  • You have a half day. Run two consecutive tracks (e.g., Track 0 L100 + Track 1 L100) with a 15-minute break. Don't run three — fatigue kills retention after 2 hours.
  • The audience is wildly mixed. Run L100 content but offer "extension challenges" for advanced participants during the hands-on. Keeps everyone engaged.

Templates and runsheets

Copyable templates live in this repo (coming soon — contributions welcome):

  • templates/pre-work-email.md
  • templates/live-session-runsheet.md
  • templates/post-work-30-day-checkin.md
  • templates/post-work-90-day-survey.md
  • templates/responsible-ai-checkpoint.md

If you've built a template that worked, please open a PR.