Product discovery workshop
Ninety minutes. A script. Names on a calendar.
A product discovery workshop is a spine, not a persona canvas. Ninety minutes, four to eight people who will sit in the interviews. One belief you are willing to have broken, who you talk to, five questions, names on a calendar. If you need a 20-person brainstorm or a 12-slide problem-space deck, this is the wrong format.
Title. Design the interviews we will actually run.
When. 10:00–11:30. Product trio plus research. Not a gallery. Walking in at :20 to “hear the questions” kills the format.
Outcomes. One testable belief a customer could prove wrong on a 30-minute call. A recruiting screen: musts, skips, a channel that is not Slack-and-hope, a named recruiter. A five-question script with follow-ups — no “would you use.” Eight to ten named conversations on a tracker, each with an outreach owner and a date this week.
Materials. Board with BELIEF / WHO / FIVE QUESTIONS / NAMES. Printed workbooks, one per chair. Paper and pens — scripts are handwritten first. Red pens for the swap-and-strike pass. Shared tracker: Name / Source / Owner / Outreach date / Interview date / Status. Timer. One laptop for the tracker; everyone else closed.
The 90-minute spine
| Clock | Min | Mode | Segment |
|---|---|---|---|
| 10:00 | 8 | Talk |
Frame the hour Not a brainstorm. Four artifacts count. Copy the belief you brought. Do not decorate it. |
| 10:08 | 15 | Activity |
The belief we need to test One sentence a customer could prove wrong. One-dot vote. One belief for two weeks. |
| 10:23 | 15 | Discussion |
Who we talk to Must / skip / how we find them by Friday. A named recruiter in the room. |
| 10:38 | 25 | Activity |
Write the five questions Last time, workaround, cost, who else, what they already spent. Strike every pitch. |
| 11:03 | 17 | Activity |
Dry-run the script Run it on a busy, slightly annoyed peer. Rewrite the question that produced an opinion. |
| 11:20 | 10 | Discussion |
Book the first ten Eight rows live. Outreach owner and date. First interviews inside seven days. |
8 + 15 + 15 + 25 + 17 + 10 = 90. That is the whole hour and a half. A beautiful canvas with zero conversations booked is a fail.
How each block earns its minutes
10:00 · Frame the hour · 8
Say. This is not a brainstorm and it is not a persona workshop. Ninety minutes. We end at the time on the board. We are here to design interviews we will actually run this week. If you leave with a beautiful canvas and zero conversations on the calendar, we failed. Four things count: a belief we are willing to have broken, who we talk to, five questions, names on a calendar. You will talk less than you want. The customer does the talking, later.
Do. Write BELIEF / WHO / FIVE QUESTIONS / NAMES ON A CALENDAR and the end time (2). Collect laptops except one scribe (2). Cross out “brainstorm” if someone says it. Visible timer: they copy the pre-work sentence into the workbook (4). 2+2+4=8. Do not start rewriting the belief yet.
On their page. In one sentence, what do we currently believe a specific person will pay us (or switch) to do. The word a customer could contradict.
10:08 · The belief we need to test · 15
Frame the test (2). Silent write six, one sentence. Each person reads in forty-five seconds — four to eight people, cut to sixty if eight, five minutes total. One-dot vote, one. Rewrite the winner on the BELIEF board in the room’s words. Strike weasel words out loud: users, better, seamless, intuitive, one. 2+6+5+1+1=15. Good: “Ops managers at 50–200 person warehouses currently stitch inventory in spreadsheets, and would switch if they could trust a weekly number without a human reconciling it.” If two beliefs remain, schedule the second for after the first eight calls. Do not merge them into mush.
10:23 · Who we talk to · 15
Draw MUST / SKIP / HOW WE FIND THEM (2). Fill from the room, eight. Force at least two disqualifiers and a channel that is not “post in Slack and hope” — warm intros, a conference list, outbound to a title (3). Assign a recruiter in the room — usually the PM — write the name on the board, two. 2+8+3+2=15. Friends and existing happy customers are last, not first. Do not start the script.
10:38 · Write the five questions · 25
You get five questions. That is a feature. Question one is context — the last time they did the job. Two through four are the workaround, the cost, and who else gets involved. Five is what they already spent — money, time, political capital. Speak the five slots once, then stop talking (3). Pairs write a script on paper, twelve. Swap with another pair for a red-pen pass: strike pitches, add “last time” language, five. Rebuild one team script on the board. Read it cold. Cut anything that sounds like a demo or a leading “so would it help if,” five. 3+12+5+5=25. If a question starts with “Would you use” or “What if we built,” throw it out. The follow-up you will actually say is usually “what happened next.”
11:03 · Dry-run the script · 17
Count off pairs. Odd person rotates in as the customer, then out (2). Visible timer: four minutes, walk and tap anyone pitching. Swap, four. Harvest: which question produced a story, which produced an opinion. Rewrite one weak question on the board, five. Last two: say the rule out loud — if you mention the product before question five, you restart the interview. 2+4+4+5+2=17. Interviewer may not explain the product. If you talk more than thirty percent of the time, you failed the round. The customer is busy and slightly annoyed, not a teammate being helpful.
11:20 · Book the first ten · 10
Open the shared sheet: Name / Source / Owner / Outreach date / Interview date / Status (1). Fill at least eight rows live from the pre-work lists, seven. Row zero: PM sends the recruiting email today — write the timestamp. Photograph the final script, two. 1+7+2=10. End standing at the tracker. End at 90 even if a ninth row is empty. First interviews happen within seven days, not “later this month.” If you cannot name eight, you shrink the screen or you admit you do not know how to find these people.
If Belief runs long, you still take Who. You steal from the next workshop’s opening, not from Booking today. A ninety-minute discovery that skips names is a homework assignment the room will skip.
What does not belong on this spine
- A persona workshop. Roles without a calendar are posters.
- A 20-person gallery. Four to eight people who will sit in the interviews.
- A “problem space” deck. One sentence on the board is the record.
- “Would you use” and “what if we built.” Those are pitches. Throw them out.
- Happy customers first. They will confirm what you already believe.
- “We’ll figure out recruiting later.” That is how the script dies in a folder.
Pre-work, in one paragraph
Twenty minutes, not a homework novel. One sentence: [specific role] at [specific company type] currently [workaround they actually do], and would switch or pay if [concrete outcome]. If it has the word “users” or “better experience,” rewrite it. Five people or firms you could actually reach in the next seven days who match that sentence — LinkedIn URLs or email, not “someone in logistics.” Forward notes from the last real customer conversation. The letter is in the workshop pre-work email guide.
After house lights: the belief, the recruiting screen, the five questions, the tracker with eight rows. Rule: if you mention the product before question five, you restart the interview. The letter is in the workshop follow-up email guide.
Places ships this hour as a kit: timed segments, say-this, do-this, the workbook prompts, the emails. Open it, put your belief in, print the paper.
Start live · Print their paper · Print this run of show (one page) · Facilitator agenda template · How to time a workshop · All guides