# Bmc Session Facilitation Design

> Designs the structure of a BMC workshop using research-based heuristics: choosing a starting point, fill-in order, session length/team composition, when the canvas isn't yet finished, evidence color-coding, and presenting the canvas to stakeholders.

- Skill: `pilot2service/bmc-session-facilitation-design` (Agent Skill)
- Install (CLI): `npx skillmds@latest add pilot2service/bmc-session-facilitation-design`
- Raw SKILL.md: https://api.skillmd.com/api/skills/pilot2service/bmc-session-facilitation-design/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- Author: Pilot2Service (https://skillmd.com/u/pilot2service)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/pilot2service/bmc-session-facilitation-design

---


# BMC Session Facilitation Design

## Purpose

Provides a research-based framework for designing a BMC workshop/session
before it starts: which block to start from, in what order to fill in the
blocks, how long the session should be and how big the team should be, when
the canvas is NOT yet finished even though the blocks are full, how to
separate evidence from assumption, and how to present the canvas to
stakeholders.

## Anchored in research

This skill is based on the owner's private research layer (an April 2026
expert interview)
(`30_domain_packs/bmc/facilitation/bmc_facilitation_heuristics.md` and
`domain_scope.md`, `status: template`, `source: research_layer`) — a
pre-filled synthesis of well-known BMC sources: Jeffries, Williams,
van der Linden, Blank (Strategyzer), Business Models Inc., and an
unattributed "YouTube synthesis" source. These files contain
`[EXPERT INPUT]`-tagged sections the owner has not yet filled in — see
"What this skill does NOT do".

## Method

1. **Choose the starting point deliberately — it isn't a neutral
   choice.** Alexander Osterwalder and Yves Pigneur's original BMC
   methodology (*Business Model Generation*, 2010) leaves the entry
   point open by design; three approaches are commonly used in practice:
   - **Customer-first** (most common): start from customer segments when
     the team's biggest uncertainty is whether a real market exists.
     Used in early-stage startups and design-thinking contexts.
   - **Value-first**: start from the value proposition when a strong
     technology or capability already exists and the question is who to
     offer it to.
   - **Current-state-first** (established organizations): honestly map
     the model AS IT STANDS TODAY before touching the future state
     (van der Linden: "the first canvas should describe reality, not the
     target state" — echoing Osterwalder's own advice to map the
     as-is model before redesigning it).
   Note (Blank): in biotech and regulated industries, IP, regulation, and
   reimbursement logic can matter more than the customer side — the order
   has to adapt to wherever the biggest risky assumptions actually sit.
2. **Fill in the right side before the left.** Segments → value
   proposition → channels → customer relationships → revenue streams →
   then key resources → key activities → key partners → cost structure.
   Rationale (van der Linden): don't build the engine before you know
   where you're driving. Exception: when an organization has locked-in
   infrastructure (patented technology, regulatory assets, a physical
   network), starting from resources is justified — but the right side
   still needs to be worked out before the canvas is internally coherent.
3. **Size the session length and team composition.** A first-round
   workshop: 2-3 hours. Under 90 minutes is insufficient. Over 4 hours
   produces over-analysis. Optimal team: 4-7 people, cross-functional —
   disagreements between sales, product, marketing, and finance surface
   hidden assumptions. A session run solo or by a single function
   produces a consensus canvas that doesn't reveal real tensions —
   Strategyzer's own facilitation guidance likewise stresses mixed
   perspectives in the room over a homogeneous group.
4. **Check readiness with the right measure, not a checklist.** The
   canvas is NOT finished just because the blocks are full ("checklist
   thinking", the "one and done" mistake — Jeffries). The canvas is
   ready when it drives decisions — revealing where the model is
   strong, where it leaks value, and which assumptions are still
   untested.
5. **Use evidence color-coding.** Green = proven, yellow = thin
   evidence, red = pure assumption — an approach popularized in the
   Business Model Canvas / Value Proposition Design facilitation
   tradition (Strategyzer, Business Models Inc.) as a lightweight way to
   separate fact from hope. Red blocks set the testing priority. For
   every assumption: what test would prove or disprove it, what signal
   would indicate pass/fail, and what's the minimum evidence threshold
   before a decision.
6. **Plan the presentation format in advance.** Revealing all nine
   blocks at once overloads the audience's cognitive capacity. Reveal
   one sticky note at a time, syncing narration to the visual. Use a
   story structure: setup (the problem), rising action (the discovery),
   climax (the model shift) — a technique borrowed from general
   presentation and storytelling craft rather than from BMC theory
   specifically, but widely used by BMC facilitators when presenting
   canvases to stakeholders and boards.

## What this skill does NOT do

- Doesn't contain the owner's own validated view on session
  facilitation — this is a research-layer synthesis, not the owner's
  own experience. Compare: `../bmc-canvas-clarity-and-iteration/SKILL.md`
  and `../bmc-antipattern-and-misunderstanding-correction/SKILL.md` ARE
  the owner's validated experience — use those as the primary source
  when available, and this skill as a supplementary framework.
- Doesn't state THIS pack's owner's own default starting point, typical
  session length, or how they personally communicate "the canvas isn't
  finished yet" to a client — these remain open in the owner's research
  notes.
- Doesn't give a fixed rule for edge cases — the research layer provides
  general guidelines, not a decision tree covering every situation.

## Refinement notes

These questions don't yet have an answer in the owner's research notes
(session 1/2 is marked "pending" in the repo's `SESSION_GUIDE.md`):

- What's your default starting point? When do you deviate from it?
- Do you follow the right-before-left order? Where do you break from it?
- What's your standard session length and best team composition, based
  on your own experience?
- What do you say to a client when they declare the canvas "finished"
  but you know it isn't?
- Do you use evidence color-coding or an equivalent system of your own?
- How do you present the canvas to stakeholders — what's your own
  narrative approach?

## Continue from here

- Next skill in the same pack:
  `../bmc-canvas-diagnostic-reading/SKILL.md` — once a session has
  produced a canvas, reading it systematically.
- Related skill in the same pack:
  `../bmc-canvas-clarity-and-iteration/SKILL.md` — the owner's own,
  validated view on variation logic and readiness criteria.
- This pack's shared guardrails: `../../CLAUDE.md`

## References

- `../../references/bmc-source-material-notes.md` — source material background
- `../../CLAUDE.md` — this pack's shared guardrails