Product Roadmap Workshop: Agenda and Facilitation

A product roadmap workshop turns what a product team knows (its strategy, its customer evidence, its delivery constraints) into a sequenced roadmap that everyone in the room can explain the next day. Run it in four parts: agree the goal and the horizon, then gather and cluster candidate problems, prioritise against criteria the group agreed before scoring, and sequence the result into Now, Next and Later with an owner per item. NN/g plans its roadmapping workshops at 2 to 4 hours; the agenda below uses the full four hours, breaks included.

For a multi-year roadmap of a whole organisation, use the Strategic Roadmap method instead, which the method library describes as spanning multiple years and covering "major initiatives, milestones, resource requirements, and dependencies across the whole organization or business unit" (90 to 180 minutes, 4 to 20 people).

What a roadmapping session has to produce

A roadmapping session has to produce a short list of problems, each tied to an objective, placed in time horizons, with a named owner for everything in the nearest horizon. Write that artefact down before you design a single block. NN/g suggests telling participants beforehand that the expected outcome is a low-fidelity roadmap.

Phrase every item as a problem. Janna Bastow of ProdPad argues that teams should prioritise problems over ideas. ProdPad's lean roadmapping guide describes each roadmap block as a problem to be solved, linked to a specific objective. A problem leaves the solution open for the team that picks it up. Write "new users abandon setup before inviting a teammate" and leave "build an onboarding wizard" for discovery.

Who to invite and what to send beforehand

Invite the people who own the inputs and the people who will build the result. That usually means product and engineering leads, design, and whoever holds the customer evidence, often someone from support or research. Decide with the sponsor who breaks a tie if the room and the sponsor disagree, and write it into the invitation.

The agenda below works for 4 to 15 people. The lower bound comes from 1-2-4-All, which needs at least four participants, and the upper bound from RICE Scoring, which the method library records for groups of 3 to 15. NN/g advises newer facilitators to start with just 3 to 6 people. With three, replace 1-2-4-All with a few minutes of silent writing and a round, and replace the dot vote with a short structured discussion, as the Dot Voting page suggests for groups under four.

Send a short pre-read. NN/g recommends distributing the high-level agenda and communicating the workshop goal and its expected outcome ahead of the session. Add the inputs the group will work from:

• the current objectives or strategy the roadmap must serve;
• the evidence you have, such as usage data, research findings and recurring support themes;
• hard constraints, including team capacity and commitments already made;
• a request to bring up to five candidate problems, written as problems.

A half-day product roadmap workshop agenda

The agenda runs four hours including two breaks: this guide's recommendation for a team that has to leave with a sequenced roadmap, and the upper end of NN/g's 2 to 4 hours. Every method block stays within the duration the method library records for that method.

Time	Minutes	Block	Method	Output
0:00–0:15	15	Open: goal, horizon, decision rule	One Word Check-In or an agenda-focused opener	Shared goal; tie-breaker named
0:15–0:35	20	Inputs	Walk through the pre-read and clear up open points	Shared facts
0:35–0:55	20	Gather candidate problems	1-2-4-All on "Which problems should this roadmap solve?"	Problem cards
0:55–1:20	25	Cluster	Affinity Map	Named themes
1:20–1:35	15	Break
1:35–1:45	10	Shortlist	Dot Voting as a filter	10 to 30 candidates
1:45–2:45	60	Score	RICE Scoring, or the Impact & Effort Matrix	Scored list with assumptions
2:45–2:55	10	Break
2:55–3:35	40	Sequence	Now, Next, Later board	Sequenced roadmap
3:35–3:50	15	Commit	Owners for each Now item, then Fist of Five	Owners; confidence check
3:50–4:00	10	Close	Parking Lot review, Check-Out Round	Parked items assigned
Total	240
	Four hours, breaks included

The opener depends on who attends. With senior stakeholders present, open with the goal and the agenda: the One Word Check-In takes 5 to 15 minutes in the library, but its "when not to use" list warns that in formal stakeholder meetings participants may read it as forced intimacy.

Clustering turns cards into themes. The Affinity Map page notes that it does not produce a ranked decision, and that under roughly 15 items a short structured discussion is faster than a full clustering exercise. The shortlist exists because the RICE Scoring page advises aiming for 10 to 30 items per scoring pass.

Scoring gets a full hour on purpose. The library's RICE entry warns that agreeing factor definitions and scoring four factors per item takes at least an hour. Detailed release planning does not fit inside this agenda: Story Mapping runs 120 to 240 minutes in the library, so book it as a separate session for the Now items. For agendas of other session types on the same spine, see the workshop agenda examples by type.

How to prioritise without letting a dot vote decide

Use the dot vote to shrink the list, and let criteria the group agreed before scoring set the priority. The Dot Voting page in the method library lists high-stakes or binding decisions under "when not to use" because a dot vote measures preference and does not weigh feasibility or evidence. The same page points to an Impact & Effort Matrix or weighted scoring when options differ on several competing dimensions, which roadmap candidates usually do. Two of the weaknesses NN/g lists are persuaded voting, where one person campaigns and others follow, and group voting, where people vote where everyone else already did.

Agree the criteria first. NN/g's approach is to have each small group propose 2 to 4 criteria, then align the whole room on 4 to 8, mixing user-oriented and business-oriented ones. With a scoring model, the group's job is to agree what each factor means for this product.

RICE scoring

Intercom's Sean McBride describes RICE as four factors for evaluating each project idea: reach, impact, confidence and effort. Each has a defined unit:

• Reach is people or events per time period, for example customers per quarter.
• Impact uses a fixed scale: 3 massive, 2 high, 1 medium, 0.5 low, 0.25 minimal.
• Confidence is a percentage, with 100% for high, 80% for medium and 50% for low.
• Effort is estimated in person-months, the work one team member can do in a month.

The method library's RICE Scoring entry gives the formula as (Reach × Impact × Confidence) ÷ Effort. McBride explains that more effort is a bad thing, so it divides the total, and that the result measures total impact per time worked. Say a hypothetical candidate reaches 400 customers per quarter at impact 1, confidence 80% and two person-months of effort: (400 × 1 × 0.8) / 2 gives 160.

The library's RICE page sets two limits. It ranks items of the same type fairly, so it does not compare fundamentally different strategic bets; score platform work and customer-facing problems in separate lists, or agree the split of capacity first. For an early-stage team with little data it recommends ICE Scoring over estimating numbers you do not have. McBride writes that RICE scores shouldn't be used as a hard and fast rule, because one project may be a dependency for another or a feature may be table stakes for certain customers. The group makes the final call under the decision rule from the opener.

Lighter alternatives

An Impact & Effort Matrix takes 30 to 60 minutes for 2 to 15 people in the library. Gamestorming's description puts high-impact, low-effort items in the best quadrant and calls low-impact, high-effort items probably best avoided. It is quicker than RICE and coarser, which suits a team without reach data. For stakeholder prioritisation the library also lists Buy a Feature, at 30 to 60 minutes for 3 to 20 people.

Sequencing: now, next, later

Sequencing places each scored problem in one of three columns. In ProdPad's definitions, Now holds the initiatives you are working on at present, Next is what happens once everything in Now is complete, and Later is everything else proposed, for an undefined future. ProdPad's guide presents the format as a replacement for Gantt timelines that trap product people into committing false dates.

Janna Bastow of ProdPad describes how she and Simon sketched the format in 2012 as Current, Near Term and Future, and Peter Gillard-Moss of Thoughtworks described Now, Next and Later columns in September 2012. Both argue against committing early: Bastow because certainty is portioned out over time horizons, Gillard-Moss because sizing the whole story list upfront forces teams to make decisions at the point when they know the least.

Run the 40-minute block in a fixed order. Fill Now first, from the top of the scored list, and stop when the owners in the room say capacity is used up. Then check dependencies: anything a Now item depends on moves into Now, whatever its score. Next and Later fill faster, because they carry less certainty and no dates.

If the Now column is bound to a fixed release date, sort its items with the MoSCoW Method once the column exists. It takes 30 to 90 minutes in the library, so add it to the separate release-planning session, or extend the sequencing block by 30 minutes and the session to four and a half hours. In the Agile Business Consortium's definition, Must Haves are the minimum usable subset the project guarantees to deliver, Could Haves the main pool of contingency, and Won't Haves what the team has agreed will not be delivered in this timeframe. Its guidance keeps Must Have effort at typically no more than 60% and Could Have effort at around 20%.

Facilitating the hard moments

Each of the three moments below has a structural answer you can prepare before the session.

A senior opinion ends the discussion

HiPPO stands for the highest paid person's opinion. During a vote, NN/g's countermeasure is to dictate the order in which each person places their votes, with junior participants first and stakeholders or subject-matter experts last. The Dot Voting page adds that strong hierarchy with public votes produces follow-the-boss voting, and that anonymous formats prevent it. Outside the vote, ask the senior person to argue for a criterion, which the whole room can then apply, instead of for a single item.

Items grow during the session

Scope creep shows up as a candidate that absorbs neighbouring cards until it is a programme. Split it back into problems on the wall, and move anything off-topic to the Parking Lot, which costs 1 to 5 minutes of session time. The library attaches two conditions: review the parked items afterwards, or the group learns that raising a concern is a dead end, and never use the parking lot to quiet a challenging participant.

The group disagrees on the criteria

When a score is disputed, ask each side to state the assumption behind its number, then record the disagreement in the confidence factor instead of averaging it away. If the argument is about the criteria themselves, stop scoring, return to the definitions agreed at the start of the block, and let the decision rule from the opener settle any tie.

After the session: the roadmap artefact and the follow-up

Within a working day, turn the wall into the roadmap: the three columns, each item phrased as a problem with its objective, the scores with their stated assumptions, and an owner per Now item. Keep it as low-fidelity as the session was. Send the parked items and anything the group agreed not to do with it, each with an owner or a decision to drop it.

Then book the review. NN/g recommends a 30-minute follow-up a month or so after the workshop so the core team can adjust the roadmap to new insights and changes. If the roadmap feeds a wider strategy cycle, the full-day strategy workshop agenda covers the session that sets the objectives, and the prioritisation workshop guide goes deeper on choosing a framework and countering seniority bias.

Your next step: write the goal sentence and the tie-break rule, copy the agenda table into your planning document, and send the pre-read with the request for up to five problem cards per person.