Running Discovery Sessions: Recruiting, Outreach, and No-Shows

Maya blocked out Tuesday and Thursday afternoons for interviews, wrote a careful email, and sent it to forty Donut CRM users who’d been active in the last month. Two replied. One booked. He didn’t show.

That’s not a bad-luck story; that’s the baseline. Cold outreach to your user base converts in low single digits, and a meaningful fraction of the people who book will no-show. If Maya’s plan for continuous interviewing depends on cold email working, the habit dies in week three — not because she stopped believing in discovery, but because the calendar stayed empty and the sprint didn’t.

The fix isn’t a better email. It’s treating recruiting as a system with multiple channels, ranked by leverage, where cold email is channel three of four — and where the machinery around the conversation (scheduling windows, reminders, overbooking) does as much work as the questions in it. This module is that system, end to end. It’s the most operational module in the course: mostly templates and playbooks, because the difference between teams that interview weekly and teams that “believe in discovery” is almost entirely logistics.

Where do interview participants actually come from?

Four channels, in descending order of leverage. Most PMs start at number three and burn out. Start at one.

1. In-product intercepts: recruit at the moment of relevance

The highest-leverage channel is your own product, because it can catch people while they’re doing the thing you want to talk about. A generic “give us feedback!” banner recruits your fans. A trigger tied to relevant behavior recruits the segment your target opportunity lives in.

Donut CRM’s target opportunity sits around what happens after meetings — follow-ups slipping, recaps going unread. So Maya’s intercept fires when a workspace processes its third meeting recap: a small dismissible card on the recap page — “You’ve had a few meetings recapped now. I’m Maya, I work on this. Can I borrow 25 minutes to hear how your after-meeting routine actually works? Pick a time →” — linking straight to her scheduling page. Third recap, not first: by then the user has real experience to narrate, and they’re demonstrably in the behavior Maya cares about.

Two rules make intercepts sustainable. Cap the exposure — show it once, dismiss forever, and cap total weekly impressions so you’re not annoying 1,400 workspaces to fill two slots. And keep the trigger tied to the current target opportunity; when the tree’s target moves, the trigger moves.

If you can’t ship an in-product surface this week, a triggered email from a real person’s address (“you hit X yesterday — can I ask you about it?”) is 80% of the value.

2. Piggyback on colleagues who already talk to customers

Somebody in your company is already in customer conversations every single day. At Donut CRM that’s Sam (founder-led sales) and Ravi (customer success and the support inbox). Their calendars are a recruiting channel that costs you nothing but a well-crafted ask.

Each function hands you a different kind of conversation:

• Sales can add you as a silent participant on late-stage calls (you hear the prospect’s world before your product reshapes it) and — more valuable and more awkward to ask for — intro you to closed-lost prospects. People who almost bought and didn’t will tell you things happy customers never will, and they have no reason to be polite.
• Customer success sees struggling accounts before the metrics do. Ravi knows which workspaces went quiet in week two. A warm intro from him (“Maya on our product team would love 25 minutes…”) converts at several times the rate of anything cold, because the relationship already exists.
• Support can tag inbound tickets that match your target opportunity. Any ticket touching follow-ups, recaps, or “what did we talk about” gets a discovery-followup tag, and Maya reviews the tagged queue weekly — each one is a pre-qualified interviewee who raised the topic themselves.

The ask has to be zero-work for your colleague, which means you write the intro for them. Here’s the message Maya sends Sam and Ravi — note the forwardable paragraph doing the heavy lifting:

Subject: 25 min/week of customer time — I'll do all the work
 
Sam / Ravi —
 
I'm running one customer interview a week to understand why
follow-ups slip after meetings (it's the strongest thing we've
found tied to week-4 retention). I need help meeting the right
people, and I've made it zero-work for you:
 
1. When you're on a call with someone who fits (small team,
   lots of meetings, struggling or churned), forward them the
   paragraph below. That's it — my scheduling link does the rest.
2. Sales: add me as a listener on late-stage or closed-lost
   calls when you think it'd be useful. I won't speak unless
   invited.
 
Two promises: I will never discuss roadmap or make commitments
on these calls, and you get my notes from every conversation
with your accounts within a day.
 
Forwardable paragraph:
---
My colleague Maya leads product here. She's talking to
customers about what happens after meetings — where follow-ups
slip and why. It's 25 minutes, no sales agenda, and it directly
shapes what we build. If you're open to it, her calendar is
here: [link]
---
 
Thanks — this is the single highest-leverage thing anyone can
do for the product this quarter.

Now the politics, honestly. Sales worries you’ll unsell — that you’ll wander into a live deal, get asked “does it do X,” answer truthfully, and cost them the close. CS worries you’ll create work — that your questions will surface complaints, raise expectations, and land as tickets in Ravi’s queue the next morning. Both fears are reasonable, and both are addressed by the two promises in that message: no roadmap commitments on the call (nothing to unsell, no expectations raised), and they get the notes (the conversation makes their job easier, not just yours). Keep both promises visibly for a month and the channel becomes self-sustaining — Sam starts forwarding people without being asked, because your notes made his next call sharper.

One more source with an asterisk: the CEO intro. When Sam personally asks a customer to talk to Maya, the meeting gets booked — nobody declines the founder. But it distorts candor. People are politer, more positive, and more careful when the boss arranged the conversation, especially at Donut’s scale where Sam probably sold them personally. Use CEO intros for logos you genuinely can’t reach any other way, and mentally discount the enthusiasm in what you hear.

3. Cold outreach to users

Cold email to your own user base works — at low conversion, with real craft required. It’s the channel you control completely, which is why it’s worth doing well, and it’s covered in depth in the next section. Just don’t let it be your only channel, because 5% reply rates mean you need a big list and a thick skin to fill one slot a week.

4. Panels and communities: last resort

Paid research panels, subreddits, LinkedIn groups. These get you people who resemble your users rather than your users, plus a professional-participant problem on panels (people optimizing for the incentive learn to say interesting things). Legitimate uses: pre-launch products with no user base, and segments you’re considering entering but don’t yet serve. For an existing product with 1,400 workspaces, if you’re on this channel, channels one and two are broken — fix those instead.

Your turn: write your channel plan

Fifteen minutes. List every person or team in your company that touches customers weekly — sales, CS, support, onboarding, community, the founder’s inbox. For each, write one line: what kind of conversation could they hand you, and what do they fear about you being in it? Then draft the internal ask (adapt Maya’s message above, forwardable paragraph included) for the two most promising, and send them today. If you’re solo on this, that’s the whole exercise; if you work as a trio, draft it together — the ask lands better co-signed by design and engineering, because it signals “the builders want to listen,” not “the PM wants requirements.”

What makes an outreach email get a reply?

The cold invite is a craft object with exactly one job: make saying yes feel small. Everything below serves that.

Subject lines. Short, specific, honest. “Can I ask you about your follow-up routine? (25 min)” beats “We’d love your feedback!” — the first is a person asking a question, the second is a company running a program. Never disguise research as something else; “Quick question” with no question inside trains people to ignore you.

Body: 90 words, ceiling. Who you are (a person with a name, not “the team”), why them specifically (name the behavior that made you pick them), what you’re asking (25 minutes, conversation, no sales agenda, no survey), and one link. Every sentence past 90 words lowers the reply rate. The specificity clause does the most work — “I noticed you’ve recapped a dozen meetings this month” tells the reader a human looked at their situation, which is rare enough to earn a reply on its own.

The ask is specific. “Pick any slot here” beats “let me know your availability” — the second creates a scheduling negotiation, and negotiations get postponed forever.

Incentives, rules of thumb. For churned users, closed-lost prospects, and busy professionals: yes, offer a gift card ($25–50 for 25 minutes) — you’re compensating people with no remaining stake in your success. For engaged active users of a product they care about: usually skip it. Many will talk because they want the product better, and leading with money can cheapen the exchange — it reframes “help shape a tool you use daily” as a $25 task, and $25-task people give $25-task answers. If actives aren’t converting without an incentive, add one; just don’t open with it. And never make the incentive contingent on “useful” feedback.

Three templates. First, the cold invite to an active user:

Subject: Can I ask you about your follow-up routine? (25 min)
 
Hi {first name},
 
I'm Maya — I lead product at Donut CRM. I noticed your team
has been using meeting recaps regularly, and I'm trying to
understand what happens after meetings: where follow-ups
happen easily and where they slip.
 
Could I borrow 25 minutes? It's a conversation, not a survey
or a sales call — I mostly want to hear recent stories from
your week. Your experience directly shapes what we build next.
 
Grab any time that works: {scheduling link}
 
Thanks either way,
Maya

Second, the churned-user invite. Different emotional register: you’re asking a favor from someone who already voted no, so acknowledge it, pay for it, and make it painless:

Subject: You left Donut CRM — can I pay you $40 to tell me why?
 
Hi {first name},
 
I'm Maya, on the product team at Donut CRM. You cancelled a
while back — totally fair, and I'm not writing to win you
back. I'm writing because people who leave see our blind
spots more clearly than anyone still here.
 
25 minutes on a call, a $40 gift card for your time, zero
sales pitch. I mostly want to hear what your routine looks
like now, after us.
 
Any slot here works: {scheduling link}
 
Either way, thanks for having given us a try.
Maya

Third, the forwardable snippet for sales and CS — kept standalone so colleagues can paste it without editing (this is the same paragraph embedded in the internal ask above; keep the two in sync):

My colleague Maya leads product at Donut CRM. She's talking
with customers about what happens after meetings — where
follow-ups slip and why. It's 25 minutes on a video call, no
sales agenda, and it directly shapes what gets built. If
you're open to it, her calendar is here: {scheduling link}

Your turn: adapt and send ten

Twenty minutes. Take the active-user cold invite, replace the behavior clause with something specific to your product (“I noticed you’ve X’d N times this month”), and the topic with your target opportunity from module 2. Pull ten users who match the behavior. Send ten individual emails — not a mail merge blast, ten emails with the specificity clause actually true for each recipient. Expect one or two replies. That’s not failure; that’s the channel’s exchange rate, and it’s why you built the other channels first.

Why do people book and then not show?

Because life happened between booking and the call — and the longer that gap, the more life happens. Show-rate is mostly a function of mechanics you control:

• Keep the booking window short: 5–7 days out, maximum. A slot booked three weeks ahead is an abstraction; a slot booked Thursday for Tuesday is a plan. Constrain your scheduling link even though it means fewer visible slots — a smaller calendar that people show up to beats a generous one they forget.
• Confirmation immediately, reminder the day before. Both from you-the-person where possible, not just the scheduling robot. The day-before note can be two sentences: “Looking forward to tomorrow at 2 — need to move it? No problem, just reply.”
• Calendar invite with the video link goes out the moment they book. A “link to follow” email is a no-show generator; at 2pm they’re searching their inbox instead of joining.
• Offer 25 minutes, not 60. People protect hours and give away half-hours. A 25-minute ask books faster, shows more, and — as module 2 argued — one well-run story gets you what you need. If it’s going brilliantly, “I could listen to this for another ten minutes — do you have them?” almost always works.

Even with clean mechanics, expect 20–40% no-shows and build for it rather than resenting it. The playbook:

Wait five minutes on the call, and spend them usefully — reviewing your discussion guide or prepping the next session, so a no-show costs you five minutes of prep time, not thirty minutes of stewing. Then one graceful same-day reschedule message, and never a guilt trip:

Subject: We missed each other — want to grab another slot?
 
Hi {first name} — no worries at all, days get away from
everyone. Still keen to hear your story if you're up for it:
{scheduling link}. If now's a bad stretch, also completely
fine — thanks either way!

One reschedule offer, then let it go. A guilt-tripped participant either ghosts forever or shows up defensive; neither helps you.

Structurally: overbook about 1.3x. If the habit needs one interview a week, book five or six per month and let attrition deliver four. Booking exactly what you need guarantees empty weeks.

And read the pattern itself as data. When Maya’s small-agency segment no-shows at twice the rate of solo founders, the lazy conclusion is “bad segment, deprioritize.” The sharper read: these people are so slammed that a booked 25-minute call loses to whatever caught fire that morning — which is evidence about their life, and suspiciously consistent with an opportunity space full of things slipping after meetings. A busy segment is not a bad segment. Sometimes the no-show rate is the interview.

What goes wrong on the call itself?

You did the hard part — a real customer is on the screen. The remaining failure modes are all self-inflicted.

Pitching instead of listening. You know the product’s story and silence is uncomfortable, so you fill it with roadmap. Every minute you talk about the product is a minute they don’t talk about their life. Rough discipline: outside the opening and the close, if you’re speaking more than 20% of the time, you’re pitching.

Leading the witness. “Don’t you find follow-ups hard to remember?” earns you a polite yes worth nothing. Module 2 covered the fix — story-based questions about specific recent past — but the habit decays fastest here, live, when a leading question feels like efficiency. It isn’t. It’s contamination.

The feature-request trap. Someone will say “you should just add X,” and the pull to respond as a product owner — evaluate, promise, deflect — is enormous. Any of those ends the discovery conversation. Take the feedback without taking on a commitment, and pivot back into their world with exact phrasing worth memorizing: “That’s really useful context — I can’t promise anything on the roadmap, but help me understand what that would let you do.” The request is data; the need underneath it is the opportunity. This is also the promise you made Sam and Ravi, so keeping it protects two things at once.

Demo-gravity. Users ask you to show them things — it’s the interaction they know, and for you it’s flattering and easy. Deflect until the last five minutes: “Happy to — let me make sure I get your story first, and we’ll leave time at the end.” A demo mid-interview converts the session into support, and you can’t un-convert it.

Recording consent. Ask at the top, every time, explicitly: “I’d like to record so I can listen instead of scribbling — the recording stays with the product team. OK?” If they hesitate, don’t record, and take notes; a slightly thinner record beats a slightly guarded participant. (Donut CRM sells meeting recording; Maya of all people asks properly.)

Two-person coverage. When the trio joins, split roles hard: one person interviews, one takes notes, and the third — if present — stays silent until invited. Three people firing questions is an interrogation panel. Tom taking notes while Maya interviews means Maya can hold eye contact and follow threads instead of typing, and it means Tom hears the customer verbatim rather than through Maya’s summary — which is half the point of doing discovery as a trio at all. Solo? Record (with consent) and snapshot immediately after; don’t try to interview and transcribe simultaneously.

How do you plan the session? The one-page discussion guide

Not a script — you’ll follow the customer’s story wherever it goes — but a one-pager that guarantees the session opens cleanly, aims at your target opportunity, and closes with the two asks people always forget. Four parts: opening (rapport and consent), story prompts tied to the target opportunity, probes to go deeper, close (intros and permission to follow up).

Here’s Donut CRM’s guide for the follow-up-slippage opportunity:

# Discussion guide — follow-up slippage (v3)
Target opportunity: "My follow-ups slip after meetings."
25 minutes. One story deeply > three stories thinly.
 
## Opening (3 min)
- Thanks; who I am; this shapes what we build, no sales agenda.
- "Nothing you say is wrong — I'm here to learn how you work."
- Recording consent (explicit yes before hitting record).
 
## Story prompts (pick one, go deep — 15 min)
- "Tell me about your most recent meeting where something
  needed to happen afterward. Start from the moment it ended."
- "Walk me through the last time a follow-up slipped —
  when did you realize, and what happened next?"
- "Tell me about the last follow-up you're really glad
  you sent. How did it come together?"
 
## Probes (use inside the story, not as new questions)
- "What did you do right then?" / "What happened next?"
- "Where were you? What else was going on?"
- "You said [their word] — say more about that?"
- Trace artifacts: sticky notes, drafts, reminders — "can
  you tell me about that note?"
 
## If they request a feature
- "Really useful context — I can't promise anything on the
  roadmap, but help me understand what that would let you do."
 
## Close (5 min)
- "Is there anything about after-meeting work I should have
  asked and didn't?"
- Ask for intros: "Who else do you know who deals with this?
  Could you forward them my calendar?"
- Permission to follow up: "As we explore ideas here, can I
  show you rough versions and get your reaction?"
- Thanks; what happens with what they shared.

The close deserves its own sentence of advocacy, because it’s what makes the habit compound. The intro ask turns every interview into a recruiting channel — warm referrals convert better than anything else on this page. And the permission-to-follow-up ask is you pre-recruiting for modules 4 and 5: when Maya has prototype concepts for the auto-drafted follow-up email, she won’t be starting recruitment from zero — she’ll have a bench of people who already said yes.

Version the guide (that v3 is real): after each interview, note which prompt opened the richest story and which fell flat, and revise. A discussion guide is a living artifact, not a form.

Your turn: write your discussion guide

Twenty-five minutes. Write your own one-pager for the target opportunity you chose in module 2, using the four-part skeleton: opening with consent, three story prompts (each one asking for a specific recent instance, never an opinion or a habit-in-general), five probes, and a close containing both asks — intros and permission to follow up. Then the test: read each story prompt aloud and check it can’t be answered with yes, no, or a preference. If it can, it’s an opinion question wearing a story costume — rewrite it. Running this as a trio? Draft prompts independently first, then merge; you’ll catch each other’s leading questions far better than you catch your own.

What you should have now

With this machinery running, the interviews stop being an event and become a pulse — one conversation a week feeding the opportunity space, indefinitely. Next: what to do with a target opportunity once the evidence is solid — generating real solution options instead of grabbing the first idea — in module 4.

Further reading

• Continuous Discovery Habits — Teresa Torres — chapters on automating recruitment are the direct ancestor of this module’s channel thinking.
• Interviewing Users — Steve Portigal — the deepest treatment of what happens inside the session, including field logistics most books skip.
• Talking to Humans — Giff Constable — short and tactical on outreach and recruiting, especially for early-stage products.