Your First 10 Customer Discovery Calls: Who to Book, How to Book Them and the Script cover image
Photo by Mina Rad on Unsplash
Back to blog
Research

Your First 10 Customer Discovery Calls: Who to Book, How to Book Them and the Script

A solo founder's plan for the first ten discovery calls: which ten people to pick, a booking message that gets replies, how many to over-book for no-shows, and a 30-minute script to read from.

Koji

Koji Team

Product · · 10 min read

You are a solo founder with an idea, a free evening and no researcher. You want to talk to ten people this fortnight and come out knowing whether the problem is real. This post covers only that job: who goes in your first ten slots, how to get them onto a calendar, and a 30-minute script you can read from. For the full method behind discovery interviews, read our reference guide, Customer Discovery Interviews: The Complete Guide.

The short version: fill the ten slots on purpose (six people who had the problem recently, two who switched or gave up, two from a neighbouring segment), book them with a message that asks for 20 minutes and names exactly what you want to learn, over-book by one or two for no-shows, and run every call on the same script so you can compare them on day 14.

Why ten, and why it matters who

Ten is not a magic number, but it is a defensible first batch. In a well-known study of 60 interviews, Guest, Bunce and Johnson found that 100 of their 109 codes, about 92%, had appeared by the twelfth interview (Field Methods, 2006). A later review of empirical saturation studies by Hennink and Kaiser put saturation at roughly 9 to 17 interviews when the group is fairly uniform and the questions are narrow (Social Science & Medicine, 2022). Ten calls with one tightly defined type of person will show you most of the big patterns. Ten calls with ten different types of person will show you ten anecdotes.

That is why the slots matter more than the count. The stakes are real: CB Insights' review of 431 VC-backed companies that shut down since 2023 found that 43% cited poor product-market fit, more than bad timing (29%) or unit economics (19%) (CB Insights, startup failure reasons). Most of those teams talked to customers. Many talked to the wrong ones, or to the right ones in the wrong way.

If you want the deeper reasoning on sample size, our guide to how many interviews are enough covers when ten is too few.

Who goes in your first ten slots

Write one sentence first: who has the problem, and when does it happen? "Ops leads at 20 to 100 person agencies, in the week they onboard a new client" is a sentence. "Small businesses" is not.

Then fill the slots like this:

SlotHow manyWhoWhy they are in the batch
Recent sufferers6Had the problem in the last 3 months, in your one-sentence segmentThey remember details, so you get facts, not opinions
Switchers1Changed tool or process for this problem in the last yearThey can tell you what pushed them and what nearly stopped them
Gave up1Tried to fix it and stoppedThey show you the cost of the workaround and where fixes fail
Neighbouring segment2Same job, one step bigger or smaller than your segmentThey tell you whether your segment is the sharp end or just a slice
Table of the first ten discovery calls: six recent sufferers, one switcher, one who gave up and two from a neighbouring segment, with why each is there.
Fill the slots on purpose so the ten calls can be compared on day 14 · Download or share

Three people do not get a slot: friends who will be kind, anyone you have already pitched, and anyone who has never had the problem but "finds it interesting". If you need help finding the six recent sufferers at all, start with our founder playbook on finding 20 people to interview when you have no audience, then come back here.

Before you book anyone, ask two or three screening questions, by message or a short form. "When did you last deal with X?" and "What do you use for it today?" are enough. If the answer to the first is "never" or "a year ago", thank them and move on. Our screener questions guide has more examples.

How to book them

Most first-time founders lose a week here. Three rules help.

Ask small and specific. Ask for 20 minutes, not "a quick chat". Say what you are trying to learn, in their words, and say you are not selling anything. Then make it easy to say yes: a booking link with three or four slots over the next five days, not an open-ended "when suits you?".

A message that works:

Hi Sam, I'm researching how agency ops leads handle onboarding a new client, especially the first week. I'm not selling anything yet; I'm trying to understand what actually happens. Would you give me 20 minutes this week? Here are a few times: [link]. If it's easier, I can send you a few questions to answer in your own time instead.

Over-book by one or two. People miss calls. MeasuringU's analysis of about 14,000 studies run through User Interviews put the average no-show rate around 9%, against 5% across MeasuringU's own 24 studies, with individual studies ranging from 0% to 34%. Their advice is to plan for a 10% to 20% no-show rate and recruit a little over (MeasuringU, typical no-show rate for moderated studies). For ten completed calls, book twelve. Our guide to reducing no-shows covers reminders and timing.

Offer an async route for the busy ones. Some of your best people will never find 20 minutes on a call. Offer them the option to answer in their own time instead. With Koji you put your script into a study and send one link: the participant talks or types with an AI interviewer that asks your questions and follows up on what they actually say, and the transcript lands next to your notes from the live calls.

Koji interview link preview card inviting people to a 5 minute voice or text conversation
What an invited person sees when you send them a Koji interview link instead of a call

Thank people for their time. For strangers, a small gift card is normal. Our guide to research participant incentives has typical amounts.

The 30-minute script

Use the same script for all ten calls. You can follow up freely, but keep the order and the core questions fixed, or you will not be able to compare calls at the end. The questions follow the rules in Rob Fitzpatrick's The Mom Test: talk about their life, ask about specific things that already happened, and talk less than they do. Our playbook The Mom Test for your next 10 customer calls goes deeper on scoring calls afterwards.

Process of a 30 minute discovery call script in six parts: open, get context, the last time story, workarounds and spend, close with asks, then write notes.
Keep the order fixed across all ten calls so you can compare them side by side · Source: Rob Fitzpatrick, The Mom Test (2013) · Download or share

Minutes 0 to 2: open. "Thanks for making time. I'm trying to understand how people like you deal with [the job]. There are no right answers, and I'm not going to pitch you anything. Is it okay if I take notes?"

Minutes 2 to 5: context. "Tell me a bit about your role and what a normal week looks like." Listen for where the problem sits in their week, if it sits there at all.

Minutes 5 to 18: the last time. "Tell me about the last time [the problem] came up. Start at the beginning." Then follow the story. "What happened next?" "Who else was involved?" "How did it end?" This is the heart of the call. Spend most of your time here and resist the urge to move on.

Minutes 18 to 25: workarounds and spend. "What do you use for this today?" "What have you tried that didn't work?" "Have you ever paid for something to fix this? What happened?" "If you had to put a cost on the last time, in hours or money, what would it be?"

Minutes 25 to 28: close with two asks. "Is there anything I should have asked and didn't?" Then: "Who else should I talk to about this?" Referrals are how your second batch of ten gets booked. If they seem keen, ask for something small that costs them a little: a second call, an intro, access to the spreadsheet they described.

Minutes 28 to 30: thank them. Tell them what happens next, if anything.

Right after: 10 minutes of notes. Write down, in their words, what they did, what it cost, what they tried, and any commitment they made. Do this before the next call or the details blur.

Questions to cut, however tempting: "Would you use a tool that...?", "How much would you pay for...?", and anything that starts with your idea. People are poor at predicting what they will do, and kind about your idea. If you want a longer list, our guide to avoiding leading questions has before and after examples.

Running the call

Talk for well under a quarter of the call. When there is a silence, count to three before you fill it; people often add the most useful detail after a pause. When they say something vague ("it's a pain"), ask for the instance: "When was the last time it was a pain? What happened?" When they ask what you are building, say you will tell them at the end, and then keep it to one sentence.

A founder on a video call with a potential customer, listening and taking notes while the other person tells their story
A founder on a video call with a potential customer, listening and taking notes while the other person tells their story

Photo by visuals on Unsplash

If you are running the calls yourself, record them with permission, or use a transcription tool, so you are not writing while they talk. If someone does the interview async, read the transcript as soon as it comes in. Interviews that stray off topic or barely answer are worth spotting early, so you can replace the person before day 14.

Day 14: read the ten calls together

Put the ten sets of notes side by side, one row per person, and answer five questions:

  1. Did at least six of the ten describe the problem unprompted, with a specific recent instance?
  2. Do the instances happen often enough to matter (weekly or monthly, not yearly)?
  3. Are people already spending time or money on a workaround?
  4. Did the switchers and the person who gave up point at the same weak spot?
  5. Did the neighbouring segment have it worse or better than your core segment?

If the answers are mostly yes, book the next ten from the referrals and go deeper. If six people described six different problems, your segment is too broad: rewrite the one-sentence description and fill the slots again. If almost nobody had a recent instance, the problem may be real but rare, which is a different business. Our guide to customer discovery vs customer validation explains what comes after this stage.

For grouping what you heard, a simple thematic analysis on a whiteboard or spreadsheet is plenty for ten calls.

When ten live calls is not the right plan

Live calls are the best way to learn at this stage, and a founder should do the first batch personally. They are slow to schedule, though, and they do not scale past a couple of dozen. When you need to hear from a second segment, a second country, or simply more people than your calendar allows, run the same script as an AI interview and send the link to everyone at once. The calls you did yourself tell you what to ask; the async interviews tell you how widespread it is.

Run your first discovery interviews with Koji

You can set up the script above in Koji in a few minutes: describe what you want to learn, and the research assistant drafts a study with your questions, which you can edit. Add your screening questions so only people who had the problem recently get through, then share one link for the people who will not book a call. Each person gets a voice or text conversation that follows up on their answers, and Koji groups what they said into themes with the quotes behind them.

A free account comes with 10 credits, and a text interview uses one, so you can run your first ten async interviews before paying anything. Start a free discovery study, or read Koji for founders first.

Run your first AI-moderated study in 10 minutes

10 free credits on signup. No credit card required.

GDPR compliantEU or US data residencyNo AI training on your data
Koji

Koji Team

Product

Share this article

Keep reading