ICP One-Pager

Draw a Boundary You Can Test

An early ideal customer profile is a claim about behavior: people on one side of a boundary should encounter similar painful work, respond through a reachable channel, recognize the same first value, and make demands that one founder can support. If those predictions fail, the profile is wrong.

That is a more useful standard than a polished description of an industry or persona. “Small businesses that need reporting” may describe a market, but it cannot tell the founder whom to interview, what workaround to expect, or which prospect to decline. A workable ICP makes those decisions before enthusiasm broadens the product again.

Use this page after choosing a market and drafting a possible wedge, but before writing outreach or scoping the first build. Complete a separate page for each plausible customer type. Evidence may eventually show that two types behave alike; do not assume it because they share a category label.

Start with the Last Occurrence

Reconstruct one recent instance of the painful work from an interview, observation, support story, or customer artifact. Write down:

• the person who owned the outcome;
• the operating environment around them: company type, team shape, process, or constraint;
• the event that made the work necessary now;
• what they actually did, including tools and people outside the official process;
• what delay, risk, rework, expense, or lost opportunity followed;
• who judged the result and who could approve a change.

Prefer a real occurrence. If none is available, label the reconstruction as a hypothesis and use it to recruit discovery conversations, not to justify a product.

Consider a modeled field-service reporting example. A twenty-person maintenance company promises clients a weekly account update. On Friday, its operations manager exports technician close-out notes, chases missing explanations, and assembles a report that must support both the client conversation and the coming invoice. The work is recurring, the deadline supplies a trigger, and the manager can judge whether a draft is usable. Those facts reveal far more than the label “field-service company.”

Now find a counterexample. A two-person residential installer may also write customer updates, but only after unusual jobs and directly from memory. An enterprise maintenance provider may have the weekly problem yet require security review, several integrations, and round-the-clock escalation. Shared industry language does not place either company inside the same useful early boundary.

Test the Four Early-Customer Conditions

The profile must survive four conditions at once.

Pain produces observable behavior. The customer already spends time, money, attention, reputation, or opportunity on the problem. Name the current alternative and the consequence it is trying to avoid. A stated preference without a recent occurrence is a lead for discovery, not evidence of fit.

The founder can reach the customer deliberately. Name a path to ten plausible people: a professional group, partner, customer list, directory, workplace, narrow search, or trusted introduction route. “Content,” “social media,” and “paid acquisition” are not reach paths until they identify where these particular customers gather and why they would respond.

Value and adoption can be judged. Identify the person who experiences the result, the person who decides whether it worked, and the person who approves payment or risk. They may be the same person. When they are not, the founder must be able to navigate the distance between them without inventing a sales and implementation team.

Trust and support fit one founder. State the data, access, onboarding, exceptions, response time, and proof the customer will reasonably require. A customer with urgent pain can still be a poor early customer if first use depends on bespoke integrations, procurement, regulated handling, constant handholding, or emergency support the founder cannot honestly provide.

Do not average these conditions into a score. A missing reach path or an impossible trust threshold can block learning even when the pain is severe.

Complete the One-Pager

Keep each answer concrete enough that another person could classify a prospect from it.

Customer boundary

• Customer: the role that owns the painful work.
• Operating environment: the company, team, process, volume, or constraint that makes the problem take this shape.
• Qualifying situation: the recurring event, deadline, handoff, obligation, or failure that creates urgency.
• Observable current behavior: the workaround, spend, labor, or avoidance already used.
• Consequence: what becomes late, risky, costly, embarrassing, or impossible when the workaround fails.

Access and decision path

• Reach path: where the founder can identify and contact the first ten plausible customers.
• Value judge: who can tell that the result improved the work, and what they can observe.
• Buyer or approver: who can authorize payment, data access, adoption, or operational risk.
• Trigger to act: why this customer would consider change now rather than tolerate another cycle.

Product and operating fit

• First value: the useful customer result, not the product event that produces it.
• Trust threshold: the minimum proof, security, privacy, accuracy, control, or recovery needed for real use.
• Support boundary: the onboarding, questions, exceptions, urgency, and response expectations one founder agrees to carry.
• Disqualifiers: observable facts that put a prospect outside this version of the profile.

Finish with one sentence:

We are studying [role] in [operating environment] who, when [qualifying situation], currently [observable behavior] and risk [consequence]. We can reach them through [path]; they experience first value when [result]; and we will initially exclude [disqualifiers].

The sentence is an index to the page, not a slogan. If it becomes vague when shortened, repair the underlying answers rather than polishing the sentence.

For the modeled example, the working profile might be operations managers at ten-to-fifty-person field-service companies that owe recurring client updates and assemble them from technician close-out notes. The reach hypothesis could be a regional service-management association and a narrow list of companies advertising maintenance contracts. First value is a reviewable Friday report that requires less reconstruction, not a successful file upload. Companies needing several source-system integrations, automatic external sending, formal enterprise procurement, or after-hours operational support sit outside the initial boundary.

Try to Disprove the Profile

Use the next ten qualified conversations to keep a small evidence record. For each person, note whether the qualifying situation occurred recently, what workaround they used, who judged the outcome, what triggered interest, how they were reached, and which trust or support demand appeared. Record disconfirming cases rather than quietly redefining the ICP after every call.

After the conversations, compare the prediction with behavior:

• Keep the profile when similar situations produce similar workarounds, value judgments, and support expectations, and the reach path yields relevant conversations.
• Narrow it when one operating condition—team size, workflow frequency, customer obligation, tool environment, or risk level—explains the mixed behavior.
• Split it when two groups share a label but differ in buyer, workflow, first value, or delivery burden enough to require different products or routes to market.
• Reject it when the pain is mostly verbal, access depends on broad attention, adoption requires an organization the founder cannot navigate, or support destroys the learning loop.

Do not broaden the profile merely because an adjacent prospect is willing to talk or pay. Ask whether serving that prospect would teach the same product, distribution, trust, and operating lessons. Revenue that pulls the founder into a different system is evidence of a different customer, not automatic validation of this one.

The page is ready when it lets the founder select interviewees, predict the current workaround, write specific outreach, name first value, anticipate the adoption path, and decline a tempting non-fit prospect. Its next job is not to become permanent. Its next job is to meet real customers and survive correction.