Sell the First Repeatable Cohort

Task

Sell the first cohort of productized-service customers.

Summary

Close enough customers under consistent terms to learn whether the offer truly repeats.

Sell the First Customers Without Rebuilding the Offer

Task ID: S2-11

The first cohort is not merely a revenue target. It is a controlled test of whether several independent customers will buy materially the same service at a real price, follow a similar sales path, and enter delivery without extensive exceptions. This article explains how to sell that cohort, preserve useful evidence, and decide whether the offer is ready to repeat.

The first cohort is a sales test and a delivery test

The offer looks repeatable on paper. It has a name, a price, a list of deliverables, and perhaps a presentation or proposal template. Then the first serious prospect asks for a different result, another buyer wants extra work included, and a third will buy only if the company changes the payment terms. Within a few conversations, the supposedly standard offer has become three custom projects.

That is the problem the first cohort must solve.

A productized service is a service with a clear customer, scope, price, delivery process, and expected result. It may contain configurable options, but the company does not redesign the work every time someone buys. Research on service productization describes standardization and modular design as ways to make an otherwise intangible service easier to explain, price, sell, measure, and deliver. The same research also recognizes that professional services may still require controlled configuration; productization does not mean pretending that every customer is identical.

The operating principle is straightforward:

Sell materially the same offer to several independent customers, record what happens from first contact through delivery, and resist changing the offer merely to rescue each sale.

The sale tests whether the customer, problem, promise, price, and buying process are credible. Delivery tests whether the scope, workload, timing, customer inputs, and expected result are realistic. A company has not proved repeatability when it can close deals only by changing the work after every conversation.

A working target of five to ten closed customers is useful as an operating target, not as a universal industry benchmark. Five customers may provide meaningful evidence for an expensive service sold to a narrow enterprise market. Ten may be inadequate for a low-priced offer sold across several industries with widely different needs. The right interpretation depends on how similar the customers are, how long the buying cycle is, how much discretion buyers have, and how much variation exists in delivery.

Assumptions. The work can begin without a formal dependency on another methodology task, but it still has practical prerequisites. The company needs a sufficiently clear draft offer, identifiable prospective buyers, the capacity to deliver what it sells, a legal agreement or statement of work, a way to invoice and collect payment, and a senior person who can make decisions when evidence challenges the offer. Without these basics, the cohort will test operational confusion rather than market demand.

What the cohort must prove

Early customers matter because they force choices that interviews and internal planning do not. Stripe’s guide to early business-to-business sales advises founders to recruit their first customers individually, using the conversations to refine positioning, learn recurring needs, and create the beginnings of references and referrals. It also recommends starting with people already connected to the company before adding carefully selected outbound prospects.

Customer involvement is valuable, but it has a cost. A study using Panel Study of Entrepreneurial Dynamics data found that involving customers early increased the likelihood of creating something customers would pay for, while also creating a risk of delaying launch; both effects were stronger for more innovative offers. The implication is not to stop listening. It is to run a bounded cohort rather than turn every prospect conversation into another open-ended design workshop.

The cohort should answer several distinct questions:

This is why the cohort should be treated as a set of explicit hypotheses. A 2024 study covering four randomized controlled trials and 759 firms found that entrepreneurs trained to use a scientific approach were more likely to terminate weak ideas and make a small number of focused changes rather than never changing or repeatedly pivoting. For this task, that means writing down what the company expects before selling:

• which customer is most likely to buy;
• which event or problem causes the customer to act;
• who has authority to approve the purchase;
• what result matters enough to fund;
• what price and terms the company expects to hold;
• which parts of the service are standard;
• which options are permitted;
• which requests will be declined or priced separately.

Those statements create a fair test. Without them, almost any outcome can be explained after the fact.

A credible cohort also contains independent buying decisions. Five departments in one parent company may produce useful delivery experience, but they do not demonstrate five independent market decisions. Likewise, five customers introduced by one unusually influential partner may validate that relationship more than they validate a general sales motion.

Choose buyers and hold the offer steady

The fastest way to make a small cohort inconclusive is to change the customer, promise, price, channel, and scope simultaneously. If the company closes two deals and loses eight, it will not know which variable mattered.

Begin with a narrow description of the buyer. A useful initial customer definition contains observable conditions rather than broad aspirations. For example:

Operations leaders at North American managed-service firms with 20 to 100 employees that have recently added a second service line and are struggling to standardize customer onboarding.

That description is more useful than “growing technology companies.” It points toward an identifiable role, a company type, a size range, a change event, and a problem.

The first prospect list should combine warm paths with targeted direct outreach. Existing customers, former customers, professional contacts, vendors, and trusted advisers can provide introductions. These conversations are efficient because some trust already exists. They must not be the only source, however. A small number of carefully researched prospects outside the immediate network helps reveal whether the offer can survive without personal goodwill. Stripe similarly recommends beginning with a founder’s network while developing new prospects by industry, role, community, association, conference, or specialist directory.

The offer should remain stable enough to make comparisons possible. “Stable” does not mean refusing all learning. It means distinguishing among three kinds of change:

1. Clarification changes how the offer is explained without changing the work.
2. Configuration selects from predefined options or modules.
3. Customization creates new work, obligations, or outcomes for one customer.

Clarifications can happen throughout the cohort. Configurations should follow written rules. Material customization should normally be rejected, priced separately, or saved for the next version of the offer.

The sales and delivery loop should look like this:

flowchart LR
    A[Offer assumptions] --> B[Qualified prospects]
    B --> C[Consistent sales conversation]
    C --> D[Paid decision]
    D --> E[Standard onboarding and delivery]
    E --> F[Review sales and delivery evidence]
    F -->|Pattern holds| G[Document and repeat]
    F -->|Pattern breaks| H[Revise one major assumption]
    H --> B

Text description: the company states its assumptions, approaches qualified prospects, uses a consistent sales conversation, closes paid customers, delivers through a common process, and reviews the combined evidence. It either documents the pattern or changes one important assumption before testing again.

Targeted outbound must also comply with the rules that apply to the sender, recipient, and channel. In Canada, commercial electronic messages generally require consent, identification, and an unsubscribe mechanism, subject to specific exceptions that must be assessed case by case. In the United States, the CAN-SPAM Act prohibits misleading headers and subject lines and requires commercial email recipients to have an opt-out method. European Union data-protection rules can apply to named business contacts, and organizations using third-party marketing lists must establish that the data was lawfully obtained and respect objections to direct marketing. United Kingdom rules distinguish between corporate subscribers and individuals such as sole traders, while data-protection duties can still apply when a named business contact’s personal data is used. These are compliance boundaries, not merely email-copy preferences, and local legal advice may be appropriate.

Run founder-led sales as a learning process

At this stage, direct participation by the founder or another senior operator is usually useful. That person understands why the offer was designed, can recognize when a prospect reveals a mistaken assumption, and has the authority to decide whether a request fits the intended service.

Founder-led selling should not mean improvised selling.

Use a common conversation structure:

Establish the situation. Ask what has changed, why the issue matters now, how the customer currently handles it, and what happens if nothing changes.

Identify the buyer and decision process. Determine who owns the problem, who controls the budget, who must approve the agreement, and whether procurement, security, legal, or finance will participate. Early sales guidance recommends directly establishing whether the contact can authorize the purchase rather than mistaking interest for buying authority.

Test the promise. Explain the expected result in the customer’s language. Ask the buyer to describe what a successful outcome would change for the business.

Present the standard offer. State what is included, what is excluded, what the customer must provide, the expected schedule, the price, and the decision required.

Handle objections without redesigning the service. Objections are evidence. A recurring objection may reveal a weak promise, missing proof, unsuitable price, or badly chosen customer. A one-off request may simply be a poor fit.

Ask for a commercial decision. The next step should be a signed agreement, deposit, purchase order, scheduled decision meeting, or clear rejection—not an indefinite sequence of “checking in.”

The company should preserve structured notes from both wins and losses. A customer relationship management system is useful if one is already in place, but a well-controlled spreadsheet or document is sufficient for the first cohort. The tool matters less than consistent fields and disciplined entry.

Separate observation from interpretation. “Buyer said the implementation appears too risky during year-end close” is evidence. “The market dislikes our onboarding” is an interpretation that may or may not hold across other prospects.

Losses are as important as wins. A closed-won-only file hides the denominator and encourages a flattering story. The company should know how many appropriate prospects were contacted, how many engaged, how many became qualified opportunities, how many received an offer, and how many made a decision.

Price for commitment and protect the scope

Interest is not the same as demand. A prospect can praise an idea, volunteer feedback, join a waiting list, or agree to a free pilot without accepting the commercial tradeoff that a real customer faces.

Research on willingness to pay illustrates the difference. A meta-analysis covering 77 studies in 47 papers, with more than 24,000 hypothetical and 20,000 real willingness-to-pay observations, found that hypothetical estimates exceeded real willingness to pay by about 21% on average. The studies largely concerned consumer goods, so that percentage should not be used as a business-to-business service benchmark. The relevant lesson is directional: stated enthusiasm is weaker evidence than an actual purchase with financial consequences.

For the first cohort, a customer should normally count as closed when a binding commercial commitment exists. Depending on the business, this may mean a signed agreement and collected deposit, an accepted purchase order, or another enforceable commitment supported by the customer’s normal payment process. A verbal yes, an unsigned proposal, or a free trial should not be counted as closed revenue.

A cohort discount can be legitimate, especially when customers are accepting a new provider, limited proof, or a more collaborative implementation. But the discount needs rules:

• state the ordinary price and the cohort price;
• record why the reduction exists;
• apply it consistently to similar buyers;
• set an expiration or customer limit;
• do not exchange a hidden discount for unlimited access, revisions, or custom work;
• record whether the buyer would have purchased at the ordinary price.

The same discipline applies to scope. A written offer should specify the result, deliverables, schedule, meetings, revision limits, customer responsibilities, response times, data or system access, acceptance conditions, and change-request process. Optional modules should have names and prices. Work outside the boundary should require a separate decision.

The history of Design Pickle provides a useful warning about promises that sound simpler than the underlying delivery model. The company launched around an “unlimited graphic design” proposition, but later acknowledged that customers could submit unlimited requests while daily production capacity remained limited. It officially removed “unlimited” from its brand on June 2, 2025, replacing it with clearer choices based on software experience, daily hours, and professional services. This is the company’s own account rather than an independent performance study, but the lesson is concrete: a memorable promise becomes dangerous when buyers reasonably interpret it more broadly than delivery can support.

The first cohort is the time to find these gaps. If buyers repeatedly misunderstand “ongoing support,” “complete implementation,” “unlimited,” “done for you,” or “guaranteed result,” the company should narrow the language before increasing sales volume.

Read the evidence without fooling yourself

The working target of five to ten customers should be treated as a starting range for gathering patterns, not a pass mark that automatically proves the offer. Small cohorts are useful for close observation, but they are statistically fragile. In a five-customer cohort, one customer represents 20% of the total. In a ten-customer cohort, one represents 10%. A single unusual account can therefore make conversion, margin, satisfaction, or delivery results look much better or worse than the underlying pattern.

A smaller cohort may be sufficient when the service has a high price, a long buying process, a narrowly defined market, and a large delivery commitment. A larger cohort may be needed when the price is low, buyers are diverse, purchases are frequent, or the company is testing several acquisition sources. Judgment should depend on information quality and customer similarity, not only count.

Track the measures that explain the count:

Do not collapse referenceability into a single checkbox. A customer may agree to take private reference calls but decline public use of its name, logo, quote, or performance data. Obtain specific permission before publishing any testimonial. Canadian guidance requires written approval and permission for covered testimonials, while U.S. guidance says endorsements must be truthful, not misleading, and disclose material relationships that could affect how the endorsement is evaluated.

Look closely at customer concentration. Research on young United Kingdom technology firms found that dependence on a key customer was negatively associated with survival, particularly for the youngest firms. Among firms that survived, key-customer dependence could also support customer-portfolio growth, showing that a major account can provide learning and credibility while still creating risk. The practical conclusion is not to reject a valuable anchor customer. It is to avoid confusing one customer’s repeated purchases, subsidiaries, or referrals with broad proof of the offer.

Basecamp’s origin illustrates what repeated customer work can eventually reveal. According to the company’s history, the founders first built an internal project-management system for their web-design work. Their clients asked to use it for their own projects, and roughly a year after public release, Basecamp generated more revenue than the design business. This example does not mean every service should become software. It shows why the first cohort’s records matter: repeated problems, common workflows, and recurring customer requests can reveal which part of a service may later support a standalone product.

The decision before moving on

The work can look complete while proving very little. Common false finishes include:

• five signed customers who bought five substantially different scopes;
• several “customers” using the service without paying;
• sales obtained only from close friends or one referral partner;
• one large customer counted as several departments;
• agreements closed through large, undocumented discounts;
• proposals that promise future custom features or uncosted integrations;
• sales notes that record wins but omit losses and disqualified accounts;
• a strong close count followed by excessive delivery labour;
• customers who purchased the founder’s personal involvement rather than the offer;
• published quotes, logos, or results without the necessary permission;
• outbound activity that ignores the recipient’s regional marketing and privacy rules.

Before the company depends on this offer for growth, it should be able to answer four questions with evidence.

Did several independent customers buy for substantially the same reason? Their wording will differ, but the underlying problem, trigger, buyer, and expected result should form a recognizable pattern.

Did they buy materially the same commercial offer? Price, scope, terms, and options do not need to be identical, but variation should follow explicit rules rather than negotiation improvisation.

Could the company deliver the promised result through a common process? The team should know the major steps, customer inputs, labour demands, direct costs, risks, and common exceptions.

Does the company know why it won and lost? Closed agreements, invoices or deposits, delivery records, and structured sales notes should support the conclusions. Assumptions and interpretations should remain visibly separate from observed facts.

If the pattern holds, the next work is to document it: refine the customer definition, sales narrative, qualification rules, proposal, pricing guardrails, statement of work, onboarding sequence, delivery checklist, and change-control process. The company can then train another person, test an additional acquisition source, or increase sales volume without redesigning the service.

If the pattern does not hold, do not disguise the result by broadening the customer definition or calling every exception a module. Choose the weakest major assumption—customer, problem, promise, price, scope, or sales path—revise it, and run another bounded cohort.

The result of this task is therefore not simply “five to ten customers.” It is a defensible answer to a harder question:

Can several customers buy and receive the same valuable service without the company starting from scratch each time?

Sources

Primary and official sources

• Stripe Atlas, “Your First 10 Customers.”
• Basecamp, “Where We Came From.”
• Design Pickle, “We Invented Unlimited Design, and Now We’re Killing It,” June 2025.
• Canadian Radio-television and Telecommunications Commission, guidance on Canada’s Anti-Spam Legislation.
• U.S. Federal Trade Commission, CAN-SPAM Act guidance.
• European Commission, guidance on business-contact data and third-party marketing lists.
• United Kingdom Information Commissioner’s Office, business marketing and data-protection guidance.
• Competition Bureau Canada, guidance on tests and testimonials.
• U.S. Federal Trade Commission, Endorsement Guides.

Open research

• Camuffo, A., et al. “A Scientific Approach to Entrepreneurial Decision-Making: Large-Scale Replication and Extension.” Strategic Management Journal, 2024.
• Newbert, S. L., Tornikoski, E. T., and Augugliaro, J. “To Get Out of the Building or Not? The Benefits and Costs of Customer Involvement During the Startup Process.” 2020.
• Shamsuzzoha, A., Blomqvist, H., and Takala, J. “Service Productisation Through Standardisation and Modularisation: An Exploratory Case Study.” 2023.
• Schmidt, J., and Bijmolt, T. H. A. “Accurately Measuring Willingness to Pay for Consumer Goods: A Meta-Analysis of the Hypothetical Bias.” Journal of the Academy of Marketing Science, 2020.
• Yli-Renko, H., Denoo, L., and Janakiraman, R. “A Knowledge-Based View of Managing Dependence on a Key Customer: Survival and Growth Outcomes for Young Firms.” Journal of Business Venturing, 2020.