Facilitate design workshop

Facilitate Design Workshop

Turn a group of people with different agendas into one decision, one map, or one ranked list — inside a timebox, with every voice used and no voice dominating.

When to use / when NOT to use

Use when:

• A project needs group alignment or group generation: kickoff, ideation, journey mapping with stakeholders, prioritization, structured critique.
• You must design the agenda, pick activities, timebox them, and handle the room.
• You're prepping a collaborative exercise or whiteboard challenge for a PD interview loop.

NOT this skill:

• Judging design quality or wording a verdict → craft-critique (this skill runs the critique session; that skill owns the feedback standard).
• Producing HMW questions and need statements from research → synthesize-research-data owns those as its Define-stage output; this skill only uses them as ideation seeds.
• Building the journey/empathy map artifact and its research grounding → map-customer-journey; this skill owns running the mapping session.
• Your own solo diverge-converge design loop → explore-divergent-concepts.
• One-on-one discovery with a stakeholder → conduct-stakeholder-interviews.
• Presenting finished work for buy-in → present-and-defend-work.

Step 0 — Should this be a workshop at all?

A workshop spends N people × H hours. Gate it:

You need…	Run instead
Information individuals already have	Interviews or a survey — cheaper, less groupthink
A decision one person owns	A decision memo with a recommendation (write-decision-rationale)
Feedback on finished work	A review, not a workshop (craft-critique)
Genuine co-creation, alignment, or a decision the group must own	A workshop. Continue.

If you can't name the artifact the group will produce, don't book the room.

The method

1. Pick the workshop type (NN/g's five)

Trigger	Type	Core output
Project starting; goals/scope/roles unclear	Discovery (kickoff)	Shared problem statement, goals, risks, roles
Team doesn't share an understanding of users	Empathy	Journey map / empathy map, prioritized pain points
Problem known; solutions needed	Design (ideation / design studio)	Clustered concepts + 1–3 chosen directions
Too many options, no agreed order	Prioritization	Ranked or quadrant-placed list with rationale
Work in progress needs structured feedback	Critique	Actionable, objective-anchored feedback list

Mixed needs → split into separate sessions. A kickoff that also ideates and prioritizes produces three shallow outputs instead of one deep one.

2. Write the workshop goal as a fill-in

"By the end, the group will have produced [artifact] so that [decision or next step] can happen on [date]."

If the blank after "so that" is vague, the workshop is a meeting wearing a costume. Fix the goal before the agenda.

3. Design the agenda — the universal skeleton

Every workshop is this shape; only the middle activities change:

1. Open        (5–10%)  — goal, ground rules, warm-up
2. Ground      (10–15%) — shared context: research insights, problem statement, constraints
3. DIVERGE     (25–35%) — individual, silent generation FIRST; quantity over quality
4. Cluster     (10–15%) — affinity-group the output; name the clusters
5. CONVERGE    (15–20%) — vote / rank / place on matrix
6. Decide      (10%)    — say the decision out loud; test for consent
7. Close       (5–10%)  — owners, next steps, where the artifact lives

Diverge-converge is the law. Never let a group discuss its way to ideas — individuals generate silently, then the group converges. Group-first generation anchors on the loudest voice and the first idea.

Timeboxing rules:

• Every activity gets a visible countdown timer. Announce it, run it, end on it.
• Generation blocks stay short — 5–8 minutes. Pressure produces volume; comfort produces polish on one idea.
• Debrief time ≥ activity time. Sketches nobody discusses are theater.
• Build ~20% buffer into the total. A workshop that fills 100% of its slot ends mid-decision.
• When a block overruns: cut a later activity, never compress the converge/decide blocks. Deciding is the point.

4. Pick activities per phase

Phase	Default activity	Mechanics
Warm-up	30-second intro + one-word check-in	Gets every voice in the room within 5 minutes — people who haven't spoken by minute 10 rarely speak at all
Ground	Insight walk-through or lightning talk	10 min max, facilitator or researcher presents; no debate yet
Diverge (ideas)	Crazy 8's: fold paper into 8, sketch 8 ideas in 8 minutes	Fat markers, no words-only sketches, no erasing
Diverge (ideas, deeper)	Design studio: sketch → 2-min pitch each → group critique → re-sketch	2 rounds; second round steals shamelessly from round one
Diverge (problems/insights)	Silent sticky-note storm, one idea per note	5 min, then read-aloud posting so ideas transfer without debate
Cluster	Affinity diagramming	Group similar notes; name clusters with a verb phrase, not a category noun ("can't tell if booking worked" not "feedback")
Converge	Dot voting (rules below)	Fast, democratic, good up to ~25 items
Converge (ties / high stakes)	Forced ranking	Each person ranks top 5; sum ranks; discuss only the disagreements
Decide	Impact-effort matrix placed live by the group	Quadrants: quick wins (do), big bets (plan), fill-ins (maybe), money pits (kill)

Dot-voting mechanics (do these exactly):

1. Clarify items before voting — questions first, campaigning never.
2. Votes per person = items ÷ 4, minimum 3, maximum 6.
3. Vote silently and simultaneously (remote: hide others' dots until reveal, if the tool allows).
4. Most senior person votes last — or blind. A HiPPO's early dots magnetize everyone else's.
5. Stacking allowed (multiple dots on one item) unless you need spread.
6. Read the result as a signal, not a verdict: top 3–5 advance to discussion; the decider still decides.

Choosing a prioritization method:

• Group in the room, < 25 items, need speed → dot voting → impact-effort matrix.
• You have real reach/effort data → RICE (score async, discuss outliers live).
• Deadline fixed, scope negotiable → MoSCoW.
• Deciding which features delight vs. merely satisfy → Kano (needs user survey data — don't fake it in the room).
• No data at all → impact-effort with explicit "these placements are hypotheses" framing.

5. Run the room

• Facilitator ≠ participant. You steward the process; you don't pitch ideas. If you must contribute, say so explicitly and hand the pen back fast.
• Instructions twice: say them AND show them written on the board/slide. Then ask one person to repeat the task back.
• Silence is working time. Don't fill silent generation with talk.
• Park it publicly: a visible "parking lot" for off-topic-but-real issues. Naming it there is what lets you move on without dismissing the person.
• Close every loop: end with decisions restated, owners named, and where the artifact lives. Send the artifact within 24h — an unshared workshop output rots into "that meeting we had."

6. Critique sessions specifically

Run critique as a facilitated format, not open-mic reactions:

1. Presenter states: what stage the work is at, what feedback they want, what's out of scope. Write these on the board.
2. Feedback must anchor to a stated objective — facilitate with "which objective does that serve?" when comments drift to taste.
3. No live redesigning. "What if you just…" goes to the parking lot.
4. Feedback standard, verdict language, and evidence discipline: per craft-critique. This skill only runs the session around it.
5. Close the loop: presenter states what they'll act on; follow up when they do — that's what keeps people bringing work to critique.

Remote facilitation deltas

• Cap at 2 hours with a break every 45–50 min. A full-day in-person workshop becomes 2–3 remote sessions.
• Pre-build the board (FigJam/Miro/Mural): one frame per activity, instructions written inside each frame, sticky templates ready. Participants should never create structure, only fill it.
• Tech check + pre-work async: context reading and even silent idea generation happen before the call; live time is for cluster-converge-decide.
• Co-facilitator required at 6+ participants: they watch chat, fix tech, and note parking-lot items while you run the room.
• Name-call gently: remote silence hides; round-robin ("Priya, then Sam") replaces "any thoughts?"
• One activity per frame; zoom the group to it. Free-roaming participants get lost and quietly disengage.
• Multitasking is the default remote failure: short blocks, frequent hand-offs, and visible timers are the countermeasure — not pleading for cameras.

Difficult-participant playbook

Pattern	What it looks like	Do this
Dominator	Answers first, longest, every time	Structured turns: "Let's hear three voices we haven't yet." Give them a job (timekeeper, scribe)
HiPPO	Senior voice; room reorients to their opinion	Brief them beforehand to speak last; silent + simultaneous voting; ask them to frame constraints, not solutions
Skeptic ("workshops are theater")	Arms crossed, "we already know the answer"	Assign them the critic role explicitly in converge — skepticism is useful there; useless in diverge
Derailer	Real issue, wrong meeting	Park it visibly, name a follow-up owner, return to the timer
Silent one	Contributes nothing aloud	Silent-writing activities are your equalizer; then read-alouds and round-robins. Never cold-spotlight with "you've been quiet"
Wordsmither	Polishes sticky-note grammar while the timer runs	"Rough is right — we fix words after we pick ideas"
Live redesigner (critique)	"What if you moved that button…"	"Parking lot — critique names problems; the presenter owns solutions"

Escalation rule: correct the process publicly ("let's return to the exercise"), correct the person privately at the break. Never argue content with a derailer mid-session — you become a participant and lose the room.

Worked example — 90-min remote prioritization workshop

Context: A booking product has a backlog of 18 proposed fixes for appointment no-shows. Team of 7 (PM, 2 engineers, designer, owner, 2 support staff). Goal fill-in: "By the end, the group will have produced a top-5 ranked fix list with impact-effort placement so that the next sprint can be scoped on Thursday."

Time	Block	Activity	Output
0:00–0:05	Open	Goal + ground rules + one-word check-in	Every voice heard once
0:05–0:15	Ground	PM lightning talk: no-show data + top 3 support themes (pre-read sent Monday)	Shared context
0:15–0:25	Diverge	Silent pass: each person adds missing fixes as stickies (pre-built frame)	18 → ~24 items
0:25–0:35	Cluster	Affinity-group duplicates; verb-phrase cluster names	~14 distinct items
0:35–0:40	Clarify	Questions only — no advocacy	Items understood
0:40–0:48	Converge	Dot vote: 4 dots each, silent, owner votes last	Top 6 surface
0:48–1:05	Decide	Place top 6 on impact-effort matrix live; engineers anchor effort, support anchors impact	Quadrant placement
1:05–1:15	Decide	Discuss disagreements only (items where placement was contested)	Top 5 ranked
1:15–1:25	Close	Restate ranking aloud; owner confirms; name Thursday scoping owner; parking lot reviewed	Decision + owners
1:25–1:30	Buffer	(Used by 0:48 block — matrix placement always overruns)	—

What this run catches — the two failures a naive version walks into. (1) At 0:05 the owner opened with "it's obviously the reminder SMS, let's just build that." The silent pass at 0:15 absorbs the pull to debate a favorite before anyone generates — the owner's fix lands as one sticky among ~24, not the anchor the room ranks around. Group-first discussion here would have ended the workshop at minute 10 with the loudest voice's answer. (2) The matrix block ran long (it always does). The buffer at 1:25 absorbed it and the 1:05–1:15 decide block stayed intact — because a session that clusters beautifully and never ranks produced nothing. Had it overrun further, the cut comes from Clarify or Cluster, never from the ranking.

Within 24h: board screenshot + top-5 list + parking-lot owners posted to the team channel.

Anti-patterns

Don't	Do
Group brainstorming out loud from minute one	Silent individual generation, then share
Agenda with no artifact named	Goal fill-in first; artifact drives agenda
"Any thoughts?" to a silent remote room	Round-robin by name; silent-writing first
Voting before clarifying questions	Clarify → then silent vote
Letting the exec vote first	Senior votes last or blind
Compressing the decide block to fit overrun	Cut a diverge round instead — deciding is the point
Critique as open-mic reactions	Scoped objectives + facilitated turns (per craft-critique)
Workshop output living in the facilitator's head	Artifact shared within 24h, owners named
Booking a workshop to deliver information	Send a doc; workshops are for co-creation and decisions
4-hour remote marathon	2h cap, breaks every 45–50 min, pre-work async

Output format (workshop plan)

# Workshop Plan: [name]
**Type:** discovery / empathy / design / prioritization / critique
**Goal:** By the end, the group will have produced [artifact] so that [decision/next step] by [date].
**Participants:** [names + roles + who is the decider]
**Mode:** in-person / remote — [tool]
**Pre-work:** [what, sent when]

## Agenda
| Time | Block | Activity | Instructions (verbatim) | Output |

## Materials
[Board link / frames, timers, markers, templates]

## Risks
[Named difficult-participant risks + planned countermeasures]

## After
[Artifact destination, share deadline, follow-up owner]

Sources

• 5 UX Workshops cheat sheet — https://www.nngroup.com/articles/5-ux-workshops/
• Workshop Facilitation 101 — https://www.nngroup.com/articles/workshop-facilitation-101/
• Facilitating a Design Studio Workshop — https://www.nngroup.com/articles/facilitating-design-studio-workshop/
• Foundational workshop activities (affinity, dot voting, forced ranking) — https://www.nngroup.com/articles/workshop-activities/
• Facilitating AI-Enhanced Workshops — https://www.nngroup.com/articles/facilitating-ai-workshops/
• Design Critiques: Encourage a Positive Culture — https://www.nngroup.com/articles/design-critiques/
• Following Up After a Design Critique — https://www.nngroup.com/articles/after-design-critique/
• Recovering Derailed Critiques — https://www.nngroup.com/articles/derailed-design-critiques/
• How-Might-We Questions — https://www.nngroup.com/articles/how-might-we-questions/
• Prioritization Methods (impact-effort, RICE, Kano, MoSCoW) — https://www.nngroup.com/articles/prioritization-methods/
• Prioritization Matrices — https://www.nngroup.com/articles/prioritization-matrices/
• Journey Mapping 101 — https://www.nngroup.com/articles/journey-mapping-101/

Boundaries

• craft-critique owns feedback quality, verdict language, and the evidence-discipline protocol; this skill owns the critique session — scope, turns, parking lot, follow-up.
• synthesize-research-data owns producing HMW questions and need statements (Define-stage outputs); this skill consumes them as ideation seeds.
• map-customer-journey owns the journey/empathy map artifact and its research grounding; this skill owns facilitating the group mapping session.
• explore-divergent-concepts owns Dinesh's solo research→concepts→pick loop; this skill owns the multi-person version.
• conduct-stakeholder-interviews owns 1:1 discovery; a kickoff workshop often follows those interviews.
• present-and-defend-work owns presenting finished work and handling dissent; workshops generate, presentations persuade.