Key takeaways
- →Eight to twelve members is the working range - below six a single cancellation guts the session, above fourteen you get four talkers and ten spectators.
- →Name the three decisions you are willing to change before you book the room; if you cannot name any, you are running a marketing event and should call it one.
- →Your own team should hold under a third of the airtime, which means capping internal attendees at roughly one per two customers.
- →Collect input silently and simultaneously before any discussion, or the largest customer in the room anchors every answer that follows.
- →The activities that work with senior people are ranking with a constrained budget, anonymous confidence votes and a pre-mortem question, not open discussion.
- →The board is kept alive by the thirty days after the session - a recap in 48 hours and an honest list of what you rejected and why.
A customer advisory board is a standing group of eight to twelve customers who meet two to four times a year to give you a steer on direction - not feedback on features. Run it as a working session and it changes decisions you were about to get wrong. Run it as a presentation with a discussion slot bolted on the end, and it produces polite nodding, a thank-you email and members who quietly stop attending after the second meeting.
The failure mode is nearly always the same: the vendor talks for two hours and calls the final twenty minutes 'open discussion'. Senior customers did not clear half a day to watch a roadmap deck. They came because being consulted by a company they depend on is genuinely valuable to them, and because they want to influence something before it is built rather than after.
This guide covers who belongs in the room, the pre-work that decides whether the session works, a half-day running order with real timings, the activities that get candid input out of people who are used to being sold to, how to keep your own team quiet, and the follow-up that determines whether anyone turns up next time.
What a customer advisory board is actually for
A board exists to answer questions you cannot answer alone: is this the right problem to be solving, what would make this product indispensable in three years, where are we about to spend a year building something nobody will adopt. These are judgement questions, and a room of experienced customers is useful for them for the same reason aggregated estimates often beat individual experts - independent perspectives, pooled, cancel out each other's blind spots.
It is not a support channel, a research substitute or a sales meeting with a nicer name. If you need to know whether a workflow confuses people, watch five users attempt it; a board of directors and heads of department will tell you what they believe their teams do, which is a different and less reliable thing. If your account managers are in the room mainly to strengthen relationships, you will get a session that behaves like a quarterly business review.
The clearest test is this: before you book anything, write down the three decisions you are genuinely prepared to change based on what happens in the session. If you cannot name any, you are running a marketing event. That is a legitimate thing to run - it is just not an advisory board, and calling it one burns the goodwill you would need to start a real one later.
How it differs from the formats it gets confused with
Most organisations already run three or four of these and assume the board will do the job of all of them. It will not, and trying makes the session incoherent. Decide which questions belong to the board - the ones about direction and trade-offs - and route everything else to the format that answers it properly.
The practical consequence is agenda discipline. If a member raises a bug or an account issue, take it, write it down, promise a named owner within a week, and move on. Letting one support escalation absorb twenty minutes is the most common way a half-day session ends with one topic covered instead of three.
| Format | What it is good at | Typical size | Where it goes wrong |
|---|---|---|---|
| Customer advisory board | Direction, strategy, three-year bets | 8-12 | Becomes a roadmap briefing with an expensive audience |
| User research interview | Observed behaviour on a real task | 1 | Used to answer strategy questions it cannot answer |
| Beta programme | Whether a built thing survives real use | 20-200 | Feedback arrives too late to change the shape |
| Quarterly business review | Account health and commercial follow-up | 3-8 | Advisory input gets filtered through the deal |
| User group or community event | Peer learning and enthusiasm at scale | 50-500 | Loudest advocates crowd out the ambivalent majority |
| NPS or satisfaction survey | Trend lines across the whole base | Hundreds | Tells you the score, never the reason behind it |
Choosing the eight to twelve people in the room
Eight to twelve is the working range. Below six, two cancellations leave you facilitating a conversation rather than a board. Above fourteen, per-person contribution falls away in the way it does in any large meeting - a handful of confident people carry the discussion and everyone else settles into audience mode, which is precisely the effort you are paying to avoid.
Composition matters more than logo quality. Mix segments so you are not designing for one buyer type, mix tenure so you have someone in their first six months alongside someone who has used the product for four years, and mix outcomes deliberately: include at least one customer who nearly left and one who is using you in a way you did not intend. The awkward members are the ones who tell you something you did not already believe.
Match seniority to the questions. If you want to ask about budget priorities and three-year direction, you need people who set budgets. If you invite a mix of executives and hands-on practitioners in the same room, expect the practitioners to defer - and design at least one silent, written activity so their view survives the meeting anyway.
Set terms of about two years and rotate roughly a third of the board each year. Permanent membership produces a comfortable group that has learnt your internal language and stopped noticing what is strange about your product. Rotation keeps the questions fresh and gives you a graceful exit from a member who has stopped contributing.
Recruiting members and setting the terms
Top Tips
- Invite personally and specifically. 'We are trying to decide whether to rebuild reporting or leave it alone for two years, and I want your view in the room' converts far better than a generic invitation to an advisory programme.
- Say what you will ask of them, in hours: two half-day sessions a year, roughly thirty minutes of pre-work before each, and the occasional email between meetings. Vague commitments get vague attendance.
- Be explicit that this is not a sales channel and that their account manager will not be in the room. Senior people assume otherwise, and the assumption changes what they are willing to say.
- Do not pay members in cash - it changes the relationship and creates procurement problems on their side. Access is the currency: early builds, direct time with your product and engineering leads, and a peer group they would not otherwise get.
- Agree the confidentiality position in writing before the first session, covering both directions. Members are often competitors with one another and will self-censor until they know what leaves the room.
- Name a single point of contact who owns the relationship year-round. Boards decay when the only contact is a calendar invitation twice a year from a shared inbox.
The numbers to design around
These are design constraints, not benchmarks. The airtime figure is the one facilitators find hardest to hold, and the one that changes the session most: strip your presentation to fifteen minutes of context and you will be astonished how much the room fills the space.
The pre-work that decides whether the session works
The pre-work is also your defence against the anchoring problem. Once the largest customer in the room has stated a position out loud, later contributions cluster around it, and you end up with a consensus that reflects seating order rather than judgement. Written answers collected in advance give you the unanchored distribution to compare against whatever the room converges on.
Send the questions three weeks out, not the agenda
An agenda tells people the running order; the questions tell them what to think about. 'Where would you spend our next twelve months of engineering time, and what would you stop?' gives someone a fortnight of background processing before they walk in.
Collect written answers before anyone travels
A short pre-session survey - five questions, ten minutes - does two jobs. It gives you the distribution of views before the room can anchor on one another, and it means the quietest member has already contributed before the session starts.
Read the pre-work and rewrite the agenda around it
If nine of eleven members raise integration pain unprompted, that becomes topic one and your planned pricing discussion moves to the next meeting. A board that sees its pre-work has changed the agenda immediately understands the session is real.
Brief your own attendees on what they may not do
No defending, no roadmap promises, no selling, no laptops. Give each internal attendee a job: one facilitates, one captures verbatim, the rest ask questions and listen. People without a job start filling silences.
Prepare the material you will show, then halve it
Fifteen minutes of context is enough: what changed since last time, what you did with their input, what you are stuck on. Everything else you were going to present is a handout.
Test the interactive parts the day before
Join codes, screen share, the ranking activity, the anonymous poll. Ten minutes of technical fumbling in front of eleven senior customers costs you more credibility than any single answer you give in the room.
A half-day running order that holds up
Three and a half hours is the right length for an in-person board. A full day sounds generous and produces two hours of genuine work surrounded by five hours of hospitality; senior people will also decline a full day far more often than they decline a morning.
Note where the presentation sits. It is fifteen minutes, it comes after the accountability round, and it ends on a question rather than a summary slide. If your context section is running to forty minutes in rehearsal, cut it - the material is almost always better as a document members can read afterwards.
0:00 - 0:10 Frame and re-contract
What we will decide today, what we will not, and the rule: we are here to listen, and we will not defend anything for the first two hours.
0:10 - 0:25 What we did with last time
Shipped, in progress, and explicitly rejected with reasons. The rejected list is what makes the board trust the next three hours.
0:25 - 0:40 Context, then stop
Fifteen minutes on what has changed and where you are stuck. No roadmap walkthrough. Finish with the first question on screen.
0:40 - 1:30 Topic one: the problem, silently first
Everyone writes for five minutes, responses go up together, then twenty minutes of discussion on the disagreements rather than the agreements.
1:30 - 1:45 Break
Take it on time. This is also where you have the quiet word with anyone dominating the room, and where members raise the thing they would not say in front of everyone.
1:45 - 2:30 Topic two: constrained ranking
Nine candidate investments, each member gets a budget of five votes and may spend no more than two on one item. The constraint is what produces a real priority order.
2:30 - 3:00 The pre-mortem question
'It is eighteen months from now and you have replaced us. What happened?' Written first, read out anonymously, then discussed. This is usually the most valuable half-hour of the day.
3:00 - 3:20 Read it back
Say what you heard in ten sentences, including the parts you disagree with. Ask the room to correct you. Corrections here prevent a recap that members do not recognise.
3:20 - 3:30 Commitments and next date
Three named owners, three dates, and the next session in everyone's calendar before they stand up.
Four activities that get honest input from senior people
All five share one property: everyone acts simultaneously and privately before anyone speaks publicly. That is what keeps a board from converging on the view of whoever spoke first, and it is why the interactive parts are worth the setup rather than being a gimmick. Session Flo runs the silent collection, the constrained ranking and the anonymous confidence vote from one room code, with results on the shared screen as they arrive.
Resist the temptation to add a sixth. A board session with three well-run activities and time to discuss the output beats six activities discussed at speed, and the discussion is where the reasoning behind the votes actually comes out.
Silent generation before any discussion
Five minutes of everyone writing, then all responses displayed at once. This is the nominal group technique in its simplest form, and it is the single highest-return change you can make to a board session. It removes the seating-order effect and gives you eleven independent views instead of one view with ten endorsements.
Constrained ranking rather than open prioritisation
Asking 'what matters most?' produces a list where everything matters. Give each member a fixed budget - five votes across nine options, maximum two on any one - and the trade-offs become visible. The items nobody spends on are as informative as the winner.
An anonymous confidence vote on your own plan
Show the direction you are leaning towards and ask, anonymously, how confident they are that it is right on a five-point scale. Senior customers are reluctant to tell a supplier its plan is weak in front of peers. A distribution with three ones in it is worth more than an hour of courteous discussion.
The pre-mortem question
'It is eighteen months from now, you have moved to another supplier, and you are explaining why in your board meeting. What is the reason?' Written privately, revealed anonymously, then discussed. It gives people permission to voice a risk without it reading as a threat, and it surfaces churn drivers a satisfaction survey never will.
A spectrum line on the contested question
Put a statement on screen - 'we should rebuild reporting before adding anything new' - and have everyone place themselves from strongly agree to strongly disagree at the same time. Then interview the extremes rather than the middle. Two minutes, and it locates the real disagreement immediately.
Keeping your own team from filling the room
Cap internal attendance at roughly one person per two customers, and make every one of them earn the seat with a defined role. Six product managers 'observing' will not observe; they will answer, explain, contextualise and defend, and within an hour the customers will have stopped volunteering anything that might need defending against.
The specific behaviour to ban is the reflexive response. A member says the onboarding is painful, and someone from your side explains that a new flow ships next quarter. It feels helpful and it is fatal - it converts a piece of feedback into a closed item, and it teaches the room that criticism will be met with a rebuttal rather than a question. The right reply is 'say more about where it hurts', and then silence.
Give your executive sponsor a script before the session, because seniority makes the reflex stronger. The sponsor's job is to open, to state plainly what the company is unsure about, and then to say very little for three hours. A chief executive who listens visibly for an entire session is the strongest signal available that the board is real.
Have one person do nothing but capture verbatim quotes. Paraphrased notes lose exactly the phrasing that makes a point land internally later - 'we bought you to stop doing spreadsheets and we still do spreadsheets' travels around your company in a way that 'concerns about manual work' never does.
If your team is talking for more than a third of the session, you are not running an advisory board. You are running a briefing for an unusually expensive audience.
Disagreement, and the member who takes over the room
Disagreement between members is the output, not an obstacle. When two customers want opposite things, you have learnt something about segmentation that a unanimous room would have hidden - and unanimity in a group of eleven experienced people usually means the question was too safe or the first answer anchored everyone else. Ask the disagreeing pair to explain the context that makes their answer obvious to them, and the underlying difference will surface within a few minutes.
The dominant voice is the more practical problem. It is often the largest account, the most senior person, or simply the one most comfortable talking, and the cost is not that they speak too much - it is that everyone else calibrates to them. Handle it structurally rather than socially: silent written rounds where everyone contributes at once, a turn-taking rule you announce at the start, and a habit of asking directly for the people who have not gone yet.
If it persists, use the break. 'You have the sharpest view in the room on this, and I need the two quieter people to disagree with you before we finish - can you hold back on the next round?' Framed as a request for their help it lands well, and it works far more often than any in-session technique.
Watch for the opposite failure too: a board that agrees with everything you propose. Comfortable consensus is what a group produces when dissent feels unwelcome, or when members have been in place long enough to have adopted your framing. Ask at least one question each session that you expect an uncomfortable answer to.
Running a customer advisory board remotely
Do not attempt a half-day remotely. Split it into two ninety-minute sessions a fortnight apart, one topic each, with the pre-work between them. Attendance holds up better and the quality of the second session improves because members have had two weeks to think about the first.
The hybrid version - four members in your office, seven dialling in - is the hardest format and the one most likely to happen. The rule that makes it survivable is that every contribution goes through a device, including the people sitting together in the room. If the in-person group discusses out loud while remote members watch, the remote half will contribute a fraction of what they would have and it will look like disinterest rather than what it is: a structural exclusion set up in the first ten minutes.
Remote sessions also need shorter blocks and more frequent acts of participation. Something for every member to do at least every ten minutes - a vote, a ranking, a written answer - keeps a video call from settling into the two-people-talking pattern. Put the results on the shared screen as they arrive, so the room can see itself thinking.
One thing genuinely improves remotely: written activities produce better output when everyone types on their own keyboard rather than passing sticky notes around a table. Use that. Run the generation and ranking online even for an in-person board, and keep the room time for the discussion that follows.
The thirty days after the session
The session is not what keeps a board alive; the follow-up is. Members judge whether their half-day was worth it by what arrives afterwards, and the recap that lands two days later - with their own words in it and the ranking they produced - is what converts a pleasant morning into a commitment to attend the next one.
The rejected list is the counter-intuitive part. Most companies report only the things they are doing, which reads as selective and leaves members wondering whether their point was heard at all. Publishing 'we are not doing this, and here is the reason' is what proves the board influences real decisions, because a group that only ever hears yes has no evidence that anything is being weighed.
Between meetings, keep one low-effort channel open - a shared thread or a quarterly note - and use it to ask small questions rather than to broadcast news. A board that hears from you twice a year has a relationship with a calendar invitation.
- ✓Recap out within 48 hours with verbatim quotes and the full ranking results
- ✓Every action has a named owner and a date, not a team and a quarter
- ✓An explicit 'not doing, and here is why' list alongside the commitments
- ✓Individual follow-up to any member whose issue was parked
- ✓Anything that came up as a support escalation routed and closed within a week
- ✓A short update at day 30 even when there is little progress to report
- ✓Next session date in calendars before members leave the room
- ○First agenda item of the next session is what happened to all of the above
Knowing whether the board is worth keeping
Attendance at the second and third sessions is your first real signal. Everyone attends the first one out of curiosity; the proportion who return after seeing what you did with their input tells you whether the board is functioning. If a member sends a deputy twice in a row, treat it as a resignation and have the conversation rather than waiting for a third no-show.
The harder measure is decision influence. Keep a simple log: for each session, which decisions changed because of what happened in the room, which were confirmed, and which you went ahead with despite the board's view. Three or four genuine changes a year justifies the programme. Zero over eighteen months means you are running a hospitality event, and it is better to convert it into one deliberately than to keep spending senior customers' time on a session that was never going to move anything.
Two soft signals are worth watching. Members who start bringing their own agenda items have taken ownership of the board, which is the state you want. Members who become willing references or introduce you to peers are telling you the relationship works - though the moment you start optimising for that, input quality falls away.
Finally, ask them. Two questions at the end of each session, answered anonymously: was this a good use of your time, and what would you change about how we run it. Give the room thirty seconds and the answers arrive before people have their coats on - and a board that is consulted about its own format tends to tell you the truth about everything else.
Frequently asked questions
How many customers should be on a customer advisory board?
Eight to twelve, with an invitation list of fourteen to allow for cancellations. Fewer than six and the session becomes a conversation that one absence can derail; more than fourteen and per-person contribution drops sharply, leaving a few confident voices and a silent majority. If you have more customers who deserve a seat than the room allows, run two boards split by segment rather than one oversized board - the input from each will be sharper because the questions can be more specific.
How often should a customer advisory board meet?
Two to four times a year. Twice annually is the minimum at which members retain context between sessions; more than quarterly and you run out of substantive questions and start filling time with updates. A common working pattern is two half-day sessions in person and two ninety-minute remote sessions between them, with the remote ones used for a single focused question rather than a general catch-up.
Should you pay customer advisory board members?
Cover travel and hospitality, but avoid cash payment or fees. Payment changes the relationship from peer advice to paid consultancy, creates procurement and disclosure problems on the member's side, and can make people reluctant to disagree with the organisation paying them. The currency that works is access: early builds, direct time with product and engineering leads, influence on decisions before they are made, and a peer network of people facing similar problems.
Who from your company should attend the session?
About one internal person per two customers, each with a defined role: a facilitator who stays neutral, a scribe capturing verbatim quotes, the executive sponsor, and one or two product or engineering leads who ask questions rather than answer them. Keep account managers out of the room. Their presence, however well intentioned, makes members read the session as commercial and quietly changes what they are willing to say about renewal risk.
What is the difference between a customer advisory board and a user group?
A user group is broad, self-selecting and largely about peer learning and enthusiasm - fifty to five hundred people who mostly like your product. A board is small, invited, deliberately mixed in temperament, and exists to influence decisions before they are made. Use the user group to build community and spread practice; use the board to test direction. Running the board like a user group is the fastest way to lose the members whose scepticism made them worth inviting.
How do you stop the session turning into a roadmap presentation?
Cap your own material at fifteen minutes, put the first question on screen at the end of it, and set an internal rule that nobody defends anything for the first two hours. Structure enforces this better than intention: if the agenda has customers writing, ranking and voting within the first forty minutes, there is no space for a presentation to expand into. Rehearse the context section, and if it runs over, cut it rather than borrowing minutes from the discussion.
Keep reading
- turning session output into decisions people act on — What to do with the rankings and quotes once the board has gone home.
- running constrained dot voting to reach a decision — The mechanics of the ranking activity in the second half of the session.
- handling a dominant voice in the room — For the board member whose view everyone else calibrates to.
- collecting anonymous feedback without losing trust — Why the confidence vote and pre-mortem answers should be unattributed.
- run your advisory board activities with Session Flo — Silent collection, constrained ranking and anonymous votes from one room code.