Product Discovery Workshop: Agenda, Roles and Deliverables

A product discovery workshop is a short, structured working session in which the people who own a business problem and the people who will build the solution agree on what to build first, for whom, and what evidence would prove it right. Done well, it turns a vague idea into one clear decision, a ranked list of risks and a scoped next step. That is why it belongs before any large software or AI budget is committed. If the next step is a small, bounded build, our proof-of-concept development starts from €5,000 one-off.
This guide covers what the workshop is, how it fits into the wider discovery phase, who to invite, a tested agenda, the deliverables to demand and what to do in the week after.
TL;DR
- A discovery workshop is a decision session, not a brainstorm: name one observable outcome, one decision and the evidence behind it before anyone enters the room.
- Keep the core group small: a decision-maker, a product lead, a designer and a technical partner, with experts invited only for the sessions that need them.
- A focused session can run 90 minutes in five blocks; a broader kickoff can fill a full day. Match the length to the number of decisions, not to the budget.
- Insist on written deliverables: a decision log, an opportunity map, open assumptions with owners and a recommendation for what to test next.
- Within 24 hours, share the outputs; within a week, put the biggest unknown in front of real customers.
Table of Contents
- What a product discovery workshop is, and where it sits
- Prepare: the week before
- A 90-minute agenda that produces a decision
- When you need a longer, broader session
- Deliverables: what the discovery phase should leave you with
- After the workshop: the first 24 hours and the first week
- What discovery costs, and what comes after
- Common mistakes to avoid
- Our perspective
- Ready to scope your project?
- FAQ
- Sources
What a product discovery workshop is, and where it sits
The workshop is one event inside a longer discovery phase (also called the software discovery process or project discovery). The phase is the period before development in which a team learns whether a problem is worth solving, who has it and which solution to try. The workshop is the moment where the stakeholders compress that learning into shared decisions.
One practitioner guide describes the session as short and decision-focused: the team names one observable outcome, such as more newly activated accounts completing their first useful task, then gathers dated evidence and picks a single decision to make. The point is that the workshop changes what the team investigates next and how it decides, rather than producing a wall of sticky notes.
It also helps to separate discovery from building an MVP. Public-sector design guidance from South Australia treats the output of discovery as a hypothesis about the target state of a service, for example what would happen if one user group no longer needed to visit a service centre to do a task. That hypothesis is then tested in a later phase with several possible solutions. In other words, discovery ends in something to test, not in finished software. If you want to see how the testing step looks for a founder, our guide to the MVP development process walks through testing one hypothesis first.
Who needs one
A workshop pays off when a decision is expensive to reverse. Typical triggers:
- A custom platform or internal system whose scope is still a list of wishes.
- An AI initiative where nobody can yet say which process, data and decision it should improve.
- A startup choosing between three possible first features.
- A mature company digitalizing a process that crosses several departments.
If the scope is already fixed and agreed, you probably need a delivery plan rather than a workshop.
Prepare: the week before
Most workshops succeed or fail before they start. The preparation below follows the one-week-before routine described by Anders Toxboe of Learning Loop.
Name the outcome and the decision
Write one sentence about an observable change, not a feature. “Launch a customer portal” is a feature. “Cut the number of order-status calls to support” is an outcome. Then write the one decision the session must produce, such as “which of three customer problems do we solve first?”
The facilitator should be able to answer three questions in advance: what decision are we trying to improve, what evidence do we have, and what will happen differently afterward. If any answer is blank, postpone the session.
Invite the smallest group that can decide
The recommended core is a product manager, a designer and a technical partner, with research and customer-facing expertise added where needed. A separate guide on defining roles adds one more seat that matters for business buyers: a single decision-maker who can approve scope and break ties. Without that person, the group produces opinions and the decision slips to a follow-up meeting.
Bring domain experts (finance, operations, compliance, support) only for the blocks that touch their area. Every extra person in the room adds talking time and subtracts decisions.
Collect evidence with dates and sources
Gather recent material and label it: support tickets, analytics, sales call notes, interview summaries, competitor observations. Date each item and note where it came from, so the room can tell fact from memory. Send a one-page pre-read one or two days before, and prepare the shared board and facilitation roles.
Separate the facilitator from the note-taker
One person runs the room; another captures decisions and follow-up actions. A product team guide also recommends a shared page for assumptions, observations, decisions and owners, and a completion standard: an action counts as done only when its evidence has been reviewed against a stated pass condition. That last rule keeps “we looked into it” from passing as a result.
A 90-minute agenda that produces a decision
This is the structure Learning Loop publishes for a 90-minute session. Use it when you have one decision to make.
| Block | Time | What happens | Output |
|---|---|---|---|
| 1. Frame and review evidence | 25 min | Restate the outcome, walk through the dated evidence | Shared starting facts |
| 2. Collect assumptions | 15 min | Everyone writes what they believe about customer needs and struggles | Assumption list |
| 3. Map opportunities | 20 min | Build an opportunity solution tree and a journey map | Opportunity map |
| 4. Prioritize one problem | 15 min | Score opportunities and pick one customer problem | Ranked list, one winner |
| 5. Plan the next learning step | Remainder | Assign an owner and a review date | Action with owner and date |
A few notes from running sessions like this:
- Block 1 is where most teams lose time. Resist storytelling and stick to the dated evidence you brought.
- Block 2 works best silently first, then aloud, so the loudest voice does not set the list.
- Block 3: an opportunity map is an interpretation of the evidence, not proof that a need exists. Treat it as provisional.
- Block 4: opportunity scoring weighs how important a need is against how well it is already served. Pick one problem, not three.
- Block 5: a next step without an owner and a date is a wish.
When you need a longer, broader session
For a new product with no shared vision, a single 90-minute block is too small. Airfocus proposes a six-part full-day framework: introductions and product vision (55 minutes), a collaborative elevator pitch (30), SMART success metrics (20), user personas (45), user stories and journey mapping as the largest block (180) and a wrap-up with team definition and next steps (45). That adds up to roughly six hours of working time before breaks.

A business analyst’s facilitation advice is useful for these longer formats: spend 10 to 20 minutes of the kickoff walking everyone through the schedule and goals, facilitate each meeting actively rather than letting it drift, and hold interviews and show-and-tell sessions beforehand so the workshop itself can focus on defining processes, spotting cross-department dependencies, prioritizing requirements and shaping the MVP vision.
Running it remotely
The same logic holds on video. Keep blocks shorter, put the agenda and the board link in the invite, give each participant a defined job (timekeeper, scribe, voice of the customer) and use silent writing steps so nobody can multitask through a discussion. Break a long day into two half-days rather than one marathon.
Deliverables: what the discovery phase should leave you with
The workshop is only worth what you can still use three months later. Public guidance on discovery says the phase should end with a Discovery Document shared with senior stakeholders. It contains the recommendations and hypotheses to test next, major findings from user research and technology discovery, the artifacts produced (including a service map) and an overview of the approach and who took part.
Translated into a checklist a business buyer can use:
- Problem statement and target outcome, in one observable sentence.
- Decision log: what was decided, by whom, on which evidence.
- Opportunity map and journey map, with the chosen problem highlighted.
- Assumption register: each unknown, its owner and how it will be tested.
- Prioritized scope for a first release or proof of concept.
- Technical findings: systems to integrate, data availability, constraints and risks.
- Recommendation and next step, with a rough effort and budget range.
If a document does not help someone make a funding or scoping decision, it is a slide, not a deliverable.
After the workshop: the first 24 hours and the first week
The follow-up is where discovery becomes delivery. Learning Loop’s follow-up routine is practical:
- Within 24 hours: share the draft opportunity map, the source evidence, the open assumptions, the decision log and the owner of the next research activity.
- Within the week: interview or observe customers to challenge the highest-priority unknown, and set up a regular customer-contact rhythm.
- At the next product check-in: compare what you observed with what you assumed, and decide whether to dig further.
- When a solution is proposed: plan a validation workshop before committing to build.
Watch who was most active in the room. A business analyst’s advice is to work with those people afterward on action items, which also builds relationships with client staff who are only part-time on discovery.
What discovery costs, and what comes after
Discovery is cheap next to the build it protects, and it ends in a defined choice of how to proceed. At NULLBIT the listed starting points are published: proof-of-concept development from €5,000 one-off, smaller targeted automations from €3,000 one-off, and smaller software solutions from €15,000 one-off on our software development page. Larger work runs under two engagement models described on our cooperation page: a fixed-price turnkey project at €5,000 to €60,000 one-off, or an agile time-and-materials engagement at €5,000 to €40,000 per month.

Which model fits depends on how settled the scope is after discovery. A clear, bounded scope suits a fixed price. A scope that will keep learning as it goes suits the agile approach without a fixed scope. If the goal is a sale or investment, it also helps to know what a buyer will check, which our technical due diligence checklist lays out across seven areas.
Common mistakes to avoid
- Too many people. Past the core group, every seat dilutes the decision.
- No decision-maker. The session ends with “we’ll circle back.”
- Solutions first. Teams jump to features before agreeing on the customer problem.
- Evidence without dates. Old anecdotes pass as current facts.
- No owners. Actions without a named person and review date do not happen.
- Treating the map as proof. An opportunity map guides what to test, nothing more.
Our perspective
We see discovery as the cheapest place to get a software or AI project right. Most of the expensive surprises in a build, such as a process nobody described, data that is not where people assumed or a feature nobody asked for, can be named in one well-run room.
Our approach is to treat AI as part of the core solution rather than an add-on, so the workshop should ask early which decisions in the business could be automated or supported by data, not bolt that question on at the end. We care about measurable business results, which is why we push every workshop toward one observable outcome and a written decision log.
Transparency matters just as much. A good discovery output tells you what is known, what is assumed and what the next step costs, so you can choose a fixed-price project, an agile engagement or a small proof of concept with open eyes. That is also how long-term partnerships start: with shared facts instead of a long list of promises.
Matija
Ready to scope your project?
If you already have an outcome in mind and want to turn it into a build plan, start with the engagement model that matches your scope. Our fixed-price turnkey projects run €5,000 to €60,000 one-off, and our agile engagements run €5,000 to €40,000 per month. A small proof of concept from €5,000 one-off is a way to test one hypothesis before a larger commitment. See the engagement models and choose how we work together.

FAQ
How long should a product discovery workshop take?
A focused session aimed at one decision can run 90 minutes. A broader kickoff for a new product, with personas, user stories and journey mapping, can fill most of a day, as in the six-part framework above. Choose the length by the number of decisions you need, and split long agendas into two half-days when remote.
Who should attend?
Keep the core group small: a decision-maker who can approve scope, a product lead, a designer and a technical partner. Add domain experts such as operations, finance or support only for the blocks that need them. If nobody in the room can say yes to a scope, the workshop will not end in a decision.
What is the difference between a discovery workshop and the discovery phase?
The phase is the whole period of learning before development: research, interviews, technical investigation and documentation. The workshop is one working session inside it where stakeholders align and prioritize. The phase ends in a document with findings and hypotheses to test next.
What should I have in hand when the workshop ends?
A decision log, an opportunity map with the chosen problem, an assumption register with owners, a prioritized scope and a recommended next step. Share the draft within 24 hours so nothing is lost, and schedule customer conversations within the week.
Sources
- How to Run a Product Discovery Workshop, Learning Loop
- Phase 3 Discovery, User-Centred Design Toolkit, South Australian Department of Treasury and Finance
- How to Run a Product Discovery Workshop That Turns Assumptions into Owned Next Steps, ONES
- How to facilitate a product discovery workshop, airfocus
- How to Run Discovery Workshops as a Business Analyst, BACareerMentor





