Facilitation guide
Most product discovery workshops fail before ideation starts. The room fills, coffee is poured, and the first sticky note goes up — but the team has never agreed on what problem they're actually solving, hasn't surfaced the constraints that will kill half the ideas they're about to generate, and has no shared understanding of how a decision gets made at the end. The session produces energy, a wall of notes, and a follow-up meeting. It doesn't produce a decision.
This guide covers how to prevent that. The structure below works with a whiteboard and a sharpie — or with a dedicated tool.
If you want a tool that runs the structure automatically, Bandos handles the facilitation so you can focus on the conversation.
The work that determines whether a session produces a decision or a debrief starts before anyone enters the room.
Before you arrive, write one sentence that pins the session to a specific customer, a specific goal, a specific friction, and a specific definition of success. Use the format: "For [persona] who wants to [goal], we will solve [friction] so they can [outcome]."
Bring it as a working draft, not a final answer. The team should be able to challenge it. The point isn't to have the right answer before the session — it's to give people something concrete to react to rather than a blank wall. A wrong problem statement that sparks a better one is infinitely more valuable than starting from nothing.
Budget ceiling. Technical limits. Timeline. Regulatory requirements. Anything that would automatically kill an idea needs to be on the wall before ideation begins.
Constraints stated late kill momentum — a team that spends 20 minutes developing an idea, only to hear "we can't do that because of X," loses trust in the process. State them early, name them explicitly, and leave them visible throughout the session.
Before anyone suggests anything, align on how the session ends. Will the group vote? Will leadership decide? Will you score options against explicit criteria?
There is no universally right answer. There is one universally wrong answer: not agreeing in advance. Sessions that end without a pre-agreed decision rule either get hijacked at the last moment by the most senior person, or end in a deferred decision that never gets made.
Six blocks, each with a name, a time, a clear output, and a facilitator tip drawn from what actually goes wrong in each phase.
Produces: a challenged and agreed problem statement; a visible list of hard constraints
Read the draft problem statement aloud. Spend 10 minutes amending it — not rewriting it. Capture any hard constraints (budget, timeline, tech limits, regulatory) on a separate section of the board. Leave them visible for the rest of the session.
Facilitator tip: Don't aim for a perfect problem statement in this block. Aim for a statement the group is willing to work from. If one objection keeps resurfacing, note it explicitly as an "open assumption" and move forward. The goal is not consensus — it's enough alignment to start.
Produces: a specific named customer the team agrees to design for
Define who the product is actually for — not as a market segment, but as a specific person in a specific context. What do they do, where do they get stuck, and what does success look like from their side of the table?
Facilitator tip: Don't let the team debate whether the persona is realistic. Bring 3 prepared or AI-generated options and vote on which one the company is best positioned to serve. Debating persona realism is the #1 time sink in discovery workshops — it produces conviction without evidence. A specific person you can disprove beats an abstract segment you can't.
Produces: a ranked list of jobs-to-be-done for the chosen persona
Surface the specific outcomes your persona is trying to achieve. For each, define the situation that triggers the need, what they're trying to do, and what success would look like. Then vote to identify the highest-value opportunity — the one that, if solved, would most change the persona's situation.
Facilitator tip: The best opportunities are specific and uncomfortable. "Users want a better experience" is not an opportunity — it's a complaint. "A first-time manager spends 45 minutes before every review cycle rebuilding context they already had three months ago" is an opportunity. Push the group toward specificity until the description makes someone in the room slightly uncomfortable.
Produces: a set of solution directions branching from the highest-value opportunity
Generate possible solutions to the opportunity identified in the previous block. Start with broad, directional options before getting specific — four high-level directions first, then develop the most promising one into concrete execution. Keep every idea tethered to the persona and opportunity above it.
Facilitator tip: Go broad before you go specific. Teams that start detailed converge on the first idea that sounds concrete — which is usually the most familiar one, not the best one. Spend the first 5 minutes generating directions without evaluating any of them. Evaluation comes after generation, not during it.
Produces: a single chosen direction with visible team commitment
Run the vote and converge on a direction. This is not a discussion — it's a decision. The vote result is the output of the session. If there's a strong dissent after the result, note it and move forward. A committed-to direction with a noted dissent is more actionable than an inconclusive discussion.
Facilitator tip: Run the vote simultaneously if at all possible. If everyone submits at the same time and results only reveal after all votes are in, the outcome reflects what the group actually believes — not what felt safe to say out loud. Visible sequential voting is structurally biased toward whoever votes first.
Produces: the 3–5 most important open questions before building
Stress-test the direction before the session ends. What would have to be true for this to succeed? What are the riskiest assumptions? What would kill this in production that hasn't been considered yet?
Facilitator tip: Ask for objections, not questions. "What would have to be true for this to fail?" surfaces more useful insight in 10 minutes than "what do you think?" does in an hour. The goal isn't to validate the direction — it's to surface the open questions that need to be answered before you commit.
The situations every facilitator hits — and what actually fixes them.
When the most senior person in the room shares a strong preference before the group has formed their own opinions, the session is effectively over. Everyone else adjusts to match — often without realising they're doing it.
Fix: Switch to anonymous written input before any verbal discussion. Ask everyone to write their perspective on a card or in a shared doc, collect the responses, and read them aloud without attribution. Even on a whiteboard, this resets the dynamic. No one knows whose idea is whose when the reading starts.
This is the most common session-killer in discovery workshops. Teams want to pick the "right" customer before they'll commit to a direction, but the debate about which customer is right often has no resolution inside a 90-minute session.
Fix: Timebox it to 5 minutes and vote. A "wrong" persona you can disprove is infinitely more useful than no persona. Picking the wrong one early surfaces the evidence that the choice was wrong. Debating endlessly surfaces nothing.
Every workshop hits the moment where someone's idea sounds exciting but is clearly solving a different problem. The energy picks up; the problem statement gets left behind.
Fix: Return physically to the problem statement. Point to it and ask: "Does this idea help [persona] with [friction]?" If the answer is no, the idea belongs in a different session. If the team insists it's better than the original problem, that's a signal to revise the problem statement — not to pursue an idea that doesn't connect to it.
This is always a process failure, not a time failure. The root cause is almost always one of two things: too much time spent in the early blocks, or no one tracking time while the facilitator is managing the room.
Fix: Don't run out of time — timebox each block at the start of the session and assign someone other than the facilitator to be the timekeeper. The facilitator's job is the conversation. The timekeeper's job is the clock. They cannot be the same person.
The post-workshop phase is where most good sessions die.
One person writes the output within 24 hours — or it won't happen. This is not optional and not a best practice: it is the single factor that most determines whether a discovery session produces anything. The longer the gap between session and documentation, the more the nuance evaporates from memory. Conclusions without rationale are almost impossible to defend in a planning meeting three weeks later.
The output format should be a single page with five fields. Not a deck, not a Notion database. One page that anyone can read in three minutes and understand what the session decided and why.
One-page session output
The output then needs two things: a review within 48 hours by whoever has final sign-off, and an explicit connection to the next planning cycle. A session output that exists in a document with no link to the roadmap, sprint, or backlog is a session that didn't happen.
Bandos generates a storyboard and feature brief automatically at the end of the session — if you run the workshop in Bandos, this section is already done before you leave the room.
Related: Problem statement template · Anonymous voting tool · Running client discovery sessions
The full 90-minute agenda in plain text — copy it into Notion, Confluence, or the Slack message you send your team before the session. No email required.
DISCOVERY WORKSHOP AGENDA — 90 MINUTES
────────────────────────────────────────
0–15 min PROBLEM ALIGNMENT + CONSTRAINTS
Amend the draft problem statement. Capture hard constraints visibly.
Produces: agreed problem statement; constraint list
15–30 min PERSONA DEFINITION
Vote on which specific customer to design for.
Produces: named persona with context and frustrations
30–50 min OPPORTUNITY MAPPING
Surface jobs-to-be-done. Vote on the highest-value opportunity.
Produces: ranked list; one winning opportunity
50–70 min SOLUTION IDEATION
Generate broad directions first. Develop the most promising one.
Produces: solution directions anchored to persona + opportunity
70–80 min VOTING + CONVERGENCE
Simultaneous vote. One direction chosen.
Produces: a committed-to direction with dissents noted
80–90 min PRESSURE-TEST
"What would have to be true for this to fail?"
Produces: 3–5 open questions before building
────────────────────────────────────────
POST-SESSION OUTPUT (within 24 hours, one page):
1. Chosen persona
2. Opportunity (job-to-be-done)
3. Solution direction
4. Open questions / riskiest assumptions
5. Next action — one owner, one deadlineBandos is a discovery workshop tool that handles facilitation, voting, and synthesis automatically — so you focus on the conversation, not the clock.
Run your first session free