Your first customer: choose one person and ask them to act

You can have a first userThe first person who uses the product. They may arrive without a structured relationship with the founder and may never pay. without having a customer; you can also have a very enthusiastic pilot userA person or organization that agrees to try the product for a defined period or purpose, usually with closer support and an agreement to provide access, feedback, or another form of participation. who never gets near the checkout. Both give you information, although they tell you different things, and nobody is safe until cash changes hands.

For this chapter, your first customerFor this playbook, the first person or organization that agrees to a clear next step around the product. The record states whether they tried it for free, paid a discount, joined an early cohort, or paid the full price. is somebody who agrees to try, buy, or explore the product with you. They may try it for free, at a discount, as part of an early cohortA small group invited to start using a product during the same defined period and under shared terms, so their usage and feedback can be compared., or as a paying customer; record exactly what they agreed to, because your first paid customerThe first person or organization to exchange money for the product or service. Payment shows something that praise, an interview, or free use can't: they wanted it enough to pay. is still the holy grail.

By the end, you'll have a list of ten real people, one person to contact first, a message that says exactly what you're asking them to do, and a date for following up.

Start with ten real people

The group you identified during validationThe work you do before building too much of the product to find out whether the problem is real, painful enough for people to act on, and shared by a group you can identify. gives you somewhere to look; now you need names, and ideally ten of them. Ten is enough to compare without spending three weeks building a giant list, color-coding it, and never contacting anybody.

You may already have names in:

• your validation interviews;
• the RedditA website made up of topic-based communities where people post questions, experiences, links, and discussions. threads, reviews, forums, and research sources where you found the problem being discussed;
• people who use or pay for another product that helps with part of the same problem;
• former colleagues, customers, suppliers, or professional contacts;
• communities and events where the people affected already spend time; and
• referrals from somebody who understands the problem and knows a person experiencing it now.

Each name needs a reason for being there. “Head of Operations at a growing company” describes a type of person; “[Name], whose company is hiring a second operations manager after they described the reporting backlog last month” gives you a person, a current problem, and a reason to contact them.

The evidence chooses who goes first

A warm relationship makes contact easier; it doesn't show that the person has the problem, wants to address it, or will tell you clearly what happened. If a the evidence for a stranger is stronger, I'd start with the stranger.

These are the questions to answer for each person:

Strong evidence of a painful problem can still point you to the wrong person if they need systems you can't access, support you can't provide, or a buying decision that'll leave you waiting months for an answer.

For the first customer, I'm especially interested in whether they'll try the product and tell you what happened, because you're still working out how it fits into somebody's life or day. The easiest person to reach may give you a kind yes and very little else; somebody who's already spent money or built a workaround has something real to compare your product against.

The different questions

A free trial, a discount, an early cohort, an interview, and a letter of intentA document recording that a person or organization intends to take a stated action under specified conditions. It can move an enterprise sale forward; it isn't a completed purchase or a guarantee that one will happen. each answer a different question:

When saying yes costs more than money

Even a free product asks for time and effort: somebody may need to learn it, move data, change a habit, involve a colleague, or risk wasting the only hour they had. So, free access can show whether they begin and continue, while a sign-up alone may only show curiosity.

A discount introduces money while acknowledging that the product is still early. Explain why the price is lower, what they'll receive, and when the discount ends, or you'll have no way to separate potential interest in the product from interest in the lower price.

For a business product, the person you contact may also be putting their judgment in front of a manager or colleague, which requires a level of advocacy on their part. They need enough information to explain the why behind the product; a small test with clear limits and an easy way to stop can make this feel safer for them. A champion who can't explain the why and any potential risk will struggle to move the purchase forward when they're not in charge of the budget.

For an enterpriseA large organization with several people involved in approving and buying a product. Enterprise sales usually take longer because the buyer may need legal, security, budget, and purchasing approval. product, a letter of intent may be the closest you get to payment before the buyer's company can approve the purchase. It can record the buyer's interest, proposed terms, timing, and conditions; it isn't a purchase, so keep every condition that still has to be met in your record.

Do enough by hand to answer one question

Early customers may need help that would be absurd to provide to everybody: manual setup, data cleanup, direct training, hand-produced output, or a by-hand version of a feature. A short note in your record is enough to set the limit and say what you're trying to learn:

I'll do [specific manual work] for [person] for [time or usage limit] to find out whether [what I think is true]. I'll stop when [result]. Anything they ask for beyond this goes is noted; it doesn't change what this test was for.

Mintlify has written about manually migrating and editing early customers' documentation, including work completed before the product was fully ready. The founders learned about onboardingThe work that helps a new user or customer set up the product, understand it, and reach the first intended result. and what the product needed by doing that work themselves; Mintlify was a funded team, so the amount of work they took on isn't a sensible target for one person. (Mintlify, 26 January 2024)

Five messages is the ceiling

So how do you get hold of these early users? You do some good old-fashioned outreach. Pick your poison - email, LinkedIn, DM... wherever you're more likely to get a response.

The key when approaching these people is to be persistent but not annoying. One unanswered message tells you little about the enthusiasm of the recipient because people miss messages, forget to reply, or mean to come back after a deadline. So set your sequence ceiling at five messages in total: the first invitation and four follow-ups, spaced to suit how you're contacting them and how long a decision usually takes. 48hrs apart is a good rule of thumb.

A direct no, a request to stop, no working (or legal) way to contact them, or a fact that rules the person out ends the sequence sooner; if somebody likes the idea but won't agree to a date or a next step, record that as interest without action and contact the next person.

An enterprise conversation can stay open for longer when one person owns the next step and you've agreed a next action or date.

Your first-customer record

The first-customer record keeps the ten names, the evidence behind them, and what happened after each message in one place. In the record, you'll choose one person and one backup, decide what you're asking for, and write down how much manual help you'll give before the work begins.

It also keeps the distinction between a first user, pilot or free userA person who uses the product without paying. Their usage can show whether the product works for them; it doesn't show willingness to pay., first customer, and first paid customer attached to the result, so a free trial, an interview, and a payment don't become “three customers” when you look back later.

A prompt to review your ten possible customers

Purpose: Compare ten real people using what you learned during validation, while keeping missing facts and reasons to rule somebody out visible.

When to use it: Once you've entered ten named people in the first-customer record.

I need to choose the first person to approach as a possible early customer.

My context
- Product or service: [what I'm offering now]
- Product stage: [prototype / lightweight MVP / pre-launch / live]
- Problem I've found: [the problem and the evidence that makes it worth investigating]
- Group I think has the problem: [the recognizable group]
- Current invitation options: [free trial / discounted trial / early cohort /
  interview / survey / letter of intent / other]
- What I most need to learn: [one question]
- What I can provide right now: [what I can and can't support]
- Geography or legal limits: [if relevant]

My ten possible customers

For each person, I'll provide:
- Name
- Current role, organization, or relevant situation
- Source URLs
- The recent problem or change I observed
- What they already do or spend
- What shows they can influence or make the decision
- Why they might deal with the problem now
- What suggests they may try an early product
- What suggests they may give detailed feedback
- How I can contact them and why that method is appropriate
- My relationship to them
- Missing facts and possible reasons to rule them out

[Paste all ten entries separately]

Your task
1. Review each person separately before comparing them.
2. For every person, separate verified facts, my assumptions, your
   interpretation, and missing information.
3. Look at what has happened recently, what they're already doing or spending,
   whether they can say yes, why they might act now, whether my current product
   can help, whether they're likely to try it and tell me what happened, and
   whether I've got an appropriate way to contact them.
4. Identify any reason to rule somebody out or any fact I need to check myself.
5. Compare the ten without assigning a made-up numerical score.
6. Recommend one person to approach first and one backup. Explain which evidence
   produced the order and which uncertainty could change it.
7. Recommend the invitation type that best answers my stated learning question.
8. Give me a list of facts and sources to check for the chosen person before I
   draft a message.

Rules
- Use only people and evidence I supplied.
- Don't invent or infer a name, role, employer, relationship, problem, event,
  intent, budget, ability to decide, email address, social profile, source, quotation,
  contact permission, or willingness to participate.
- A public profile or email address doesn't automatically give permission for
  every type of contact.
- Keep friendly relationships separate from evidence of demand.
- Keep interviews, surveys, free use, discounted use, letters of intent, and
  payment as different results.
- Identify missing information instead of filling it with a plausible guess.

Return exactly these headings
## Review of each person
## Facts, assumptions, and missing information
## Reasons to rule somebody out and checks required
## Comparison of the ten
## Person to approach first
## Backup person
## Recommended invitation and the question it tests
## Facts and sources to check myself

Check the person and contact details yourself

Before using the second prompt, open the sources attached to the chosen person and confirm their identity, current role, recent problem or change, and the way you plan to contact them. A model can help compare the information you supply; it can't turn a stale job title, guessed email pattern, or closed inbox into a real way to reach somebody.

A prompt to draft the first invitation

Purpose: Draft one initial invitation and up to four follow-ups from facts you've checked yourself.

When to use it: After choosing the person, checking every source, and deciding which type of invitation fits the question you need to answer.

Draft a first-customer invitation and follow-up sequence for one person. The
sequence may contain no more than five messages in total: one initial message
and up to four follow-ups.

My product and question
- Product or service: [what I'm offering now]
- Product stage: [prototype / lightweight MVP / pre-launch / live]
- Problem it addresses: [one or two sentences]
- What I need to learn: [one question]
- Invitation type: [free trial / discounted trial / early cohort / interview /
  survey / letter of intent / other]
- Exact action I'm asking for: [the next step]
- What the person will receive: [access, support, discount, terms, or other]
- Time, effort, access, or internal support this asks from them: [what saying
  yes requires]
- What they may need to explain to somebody else: [known detail, or unknown]
- What I can offer to make the first test easier to try: [support, clear limits,
  an easy way to stop, relevant proof, or another known option]
- Manual work I'll provide: [specific limit]
- Facts or claims I must avoid: [list]

Verified person context
- Name and preferred form of address: [verified]
- Current role, organization, or situation: [verified]
- Recent problem, event, or change: [verified]
- Current approach or spend: [verified, or unknown]
- Can they say yes?: [verified, or unknown]
- Source URLs: [direct links I opened]
- How I can contact them: [verified]
- Relationship to me: [warm / referred / public conversation / no relationship]
- Previous contact: [dates, messages, replies, or none]

My voice
- How I normally write: [brief description]
- Examples: [paste two or three short messages I wrote]
- Words or approaches to avoid: [list]

Your task
1. Draft a concise initial message tied to the verified context and one clear
   action.
2. State the invitation accurately, including whether it's free, discounted,
   cohort-based, paid, or an interview or survey.
   Make the time, effort, access, or internal support required clear without
   inventing a concern the person hasn't expressed. Include any supplied
   support, limit, stop point, or relevant proof that makes the test easier to
   try.
3. Draft up to four brief follow-ups. Give each message a different job; don't
   restate the first message with new wording.
4. Include a stop note beside every message: send only if [condition].
5. End the sequence after five total messages, or sooner after a direct no,
   request to stop, no working way to contact them, a fact that rules the
   person out, or a completed next action.
6. List every missing fact or personal detail I can't support before the drafts.

Rules
- Use only the verified facts and voice examples I supplied.
- Don't invent or infer familiarity, praise, urgency, intent, ability to decide, budget,
  results, quotations, contact permission, product capability, dates, or links.
- Don't disguise a sales invitation as neutral research.
- Don't add a demo link, calendar link, attachment, discount, deadline, or claim
  unless I supplied it.
- Keep the initial message to one purpose and one ask.
- Distinguish facts, assumptions, and recommendations.
- Never send, schedule, post, or upload anything.

Return exactly these headings
## Missing facts and personal details I can't support
## Recommended approach
## Message 1: initial invitation
## Message 2: follow-up
## Message 3: follow-up
## Message 4: follow-up
## Message 5: final follow-up
## Stop conditions
## Result fields to update in my record

Good enough for now

Congratulations! Your record should now contain ten real people, your first invitation has been sent, and you've recorded when to follow up. You can't control the reply; whatever comes back belongs in your record, including silence, a no, free use, a pilot, an interview, a letter of intent, or payment (fingers crossed).

The next chapter turns those real cases into your ideal customer profileA description of the customers most likely to need, buy, and get a good result from your product, based on patterns you can see across real people rather than one imagined customer.. A single customer can teach you a lot about one person; the next job is finding which parts show up time and time again, so you know who to contact next.

Sources

• Graham, Paul. “Do Things That Don't Scale.” July 2013. Accessed 11 September 2026.
• McKenzie, Patrick. “Your First 10 Customers.” Stripe Atlas. Accessed 11 September 2026.
• Y Combinator. “How to Get Your First 10 Customers.” 22 June 2026. Accessed 11 September 2026.
• Kumparak, Greg. “50 Founders Share How They Got Their First Customers.” Y Combinator, 9 August 2023. Accessed 11 September 2026.
• Wang, Han. “How We Grew Mintlify by Doing Things That Don't Scale.” Mintlify, 26 January 2024. Accessed 11 September 2026.
• Bull, Alice. “Starting Your Go to Market: Doing Things That Won't Scale.” Romy, 2 September 2026. Accessed 11 September 2026.