Interview Customers for Evidence

A customer interview is not a pitch meeting with disguised questions. It is a bounded attempt to understand a problem through the buyer's recent behavior. The most useful evidence is usually a story about what happened, who paid, what workaround was used, and what consequence followed.

People are often poor at predicting their own future purchases. They may say a service sounds useful and never buy it. They may describe a problem vividly but have no budget or authority to solve it. The interview should reduce this uncertainty by asking about the recent past rather than collecting optimistic predictions.

The goal of this chapter is to run eight useful conversations, record evidence consistently, and decide whether a candidate task deserves a paid demand test.

Define the Evidence You Need

Before recruiting anyone, write the decision the eight interviews will inform. For example:

• "Should I continue investigating a maintenance-request triage offer?"
• "Is the recurring problem actually the missing data, the response deadline, or the approval?"
• "Who has budget and authority for a small paid pilot?"
• "What workaround currently costs the buyer time or money?"
• "Which result could be made small enough to buy and verifiable enough to deliver?"

If the decision is "Should I start a business?" the interviews are too diffuse. Narrow the question to one cluster from the task map.

Recruit People Who Have the Problem

Choose people who recently experienced the workflow:

• Current or former buyers.
• Managers responsible for the output.
• Staff performing the work.
• Owners of similar businesses.
• Suppliers who see the failure between teams.

Do not rely only on friends or peers who share your enthusiasm. Include people who can pay, approve, or block a purchase.

Use an honest invitation:

> I am researching how small property offices handle maintenance requests that arrive through different channels. I am not selling anything in this call. Could I ask you about how the process worked over the last month and what created the most rework? I need about 20 minutes.

This invitation states the topic, boundary, and time. It does not promise a free audit or hide a sales agenda.

Respect contact rules and privacy. Do not scrape personal contact details or send mass messages where consent or anti-spam rules prohibit it. If speaking with staff, avoid soliciting confidential records or information they are not authorized to share.

Use a Past-Behavior Conversation

A repeatable interview can have seven sections.

1. Context

• What is your role in this process?
• What kind of customer or project were we discussing?
• How often did this type of work occur recently?

2. Last occurrence

• Tell me about the most recent time this happened.
• When was it?
• What started the work?

Specific time anchors improve memory and reduce generic answers.

3. Workflow

• Walk me through the steps.
• Who handled each step?
• Where did information come from?
• What system held the result?

4. Spend and effort

• Did anyone pay for a tool, employee, contractor, or workaround?
• Approximately how many people and hours were involved?
• What did a delay or error cost?
• What have you tried to fix it?

Use ranges. Do not pressure the person to disclose confidential financial data.

5. Consequence

• What happened when the work was late or wrong?
• Who noticed?
• What decision or customer relationship was affected?
• What made the result trustworthy?

6. Boundaries and authority

• Who approves a small change in this process?
• What privacy, safety, legal, or quality requirements apply?
• What would prevent a solution from being used?

7. Next step

Ask for one of these only when the evidence warrants it:

• A short example workflow.
• An introduction to the person who owns the budget.
• Permission to present a paid pilot.
• A referral to someone with the same problem.

Do not ask a general "Would you buy this?" question as the central evidence.

Ask Better Questions

Strong question patterns

• "What did you do the last time this happened?"
• "Who paid for it?"
• "What did you try first?"
• "What made you stop or change the workaround?"
• "What happens if the result is two days late?"
• "Who else has to approve?"
• "Can you show me a sanitized example?"
• "What would make a small test unacceptable?"

Weak question patterns

• "Wouldn't this save you time?"
• "Do you think AI could help?"
• "If I built this, would you use it?"
• "How much would you pay for a perfect solution?"
• "Is this a big problem?"

The weak questions ask the buyer to imagine a solution or give a socially pleasant answer. The strong questions recover facts.

Separate Problem Evidence from Solution Reaction

A person can confirm a problem and still reject your proposed solution. Record these separately:

• Frequency of need.
• Cost or consequence.
• Current workaround.
• Trust requirement.
• Decision authority.
• Budget timing.
• Reaction to a bounded result.
• Objections to price, process, privacy, or scope.

This separation prevents one favorable reaction from inflating confidence. It also prevents a rejection of one solution from discarding a real problem.

Record Interviews the Same Day

Use a standard note template:

• Interview ID and date.
• Buyer type and role category.
• Relevant recent event.
• Trigger.
• Steps and handoffs.
• People, time, and spend ranges.
• Failure and consequence.
• Current tool or workaround.
• Approval and compliance boundaries.
• Exact quotes, kept short and accurate.
• Evidence gained.
• Contradictions or unknowns.
• Next step and permission.

Do not invent quotes or embellish amounts later. If the person did not give a number, write "not provided." If you inferred a cost, label it as your estimate.

Use identifiers instead of unnecessary personal data

Store the minimum information required. A business name may be unnecessary. A role category and an interview date may be enough. If a person provides confidential data, do not copy it into a consumer AI tool without authorization and a lawful basis.

Listen for Contradictions

Interviews become useful when they reveal tension:

• The buyer says speed matters, but the approval process takes two weeks.
• The owner wants automation, but staff must inspect every output.
• A tool already exists, but training and adoption are poor.
• The buyer pays for a vendor, but still performs most of the work internally.
• Price is supposedly the main issue, but the buyer refuses to share data with a lower-cost provider.

Contradictions are not necessarily deception. They often show that the real problem differs from the stated one. Record them and ask follow-up questions.

Evaluate the Eight Interviews

Use four evidence levels for each buyer:

1. No active problem: The workflow is infrequent, inexpensive, or fully solved.
2. Pain without purchase path: A real consequence exists, but no authority or budget is visible.
3. Qualified opportunity: Recent spend, recurring trigger, material consequence, and a reachable decision process are visible.
4. Test-ready: A qualified opportunity accepts a bounded next step such as a paid pilot discussion or approved review of a small offer.

Do not average the levels into a meaningless score. Count them.

Continue when:

• At least a few interviews show recent spend or serious workaround cost.
• The same trigger and consequence repeat.
• You can identify who approves a small test.
• You can define a safe, verifiable result.

Change when:

• Buyers experience different problems under one broad label.
• Your proposed result solves a symptom but not the expensive consequence.
• The buyer group is reachable, but the decision process is out of reach.

Pause when:

• The urgent issue is personal cash, health, or safety.
• You cannot obtain access without using prohibited data.
• Every route depends on a licence or professional judgment you do not have.

A Worked Interview

An illustrative buyer runs a small cleaning company. She says customer questions are overwhelming. A weak interviewer immediately presents an AI inbox assistant. A stronger interviewer asks about the last week.

The buyer describes 18 inquiries arriving through phone, a website form, and a marketplace app. A supervisor copies them into one notebook, while the owner quotes larger jobs. Two requests were missed, and one customer chose another provider. The company pays for the marketplace app and part of the supervisor's time. The owner controls new software, but another manager must approve data handling.

The expensive consequence is not typing replies. It is missing a quote request and failing to prove who responded. A small pilot might organize and flag requests rather than answer every inquiry. The buyer says she would review a one-page proposal, which is a test-ready next step but not a sale.

This is a composite example. It does not claim a real business purchased a service.

Source and Decision Caution

Interview evidence is not statistically representative merely because eight people participated. It is qualitative evidence about mechanisms and willingness to take a next step. State the sample honestly. Consent, recording, privacy, and employment rules vary by jurisdiction, so confirm the requirements that apply to your interviews.

Do not request sensitive medical, financial, identity, employee, or customer information unless it is necessary, authorized, and handled securely. If a workflow touches children, vulnerable adults, regulated advice, legal rights, health, or safety, verify professional and safeguarding requirements before offering a result.

AI transcription can mis-hear names, amounts, and commitments. Review recordings or notes before quoting. Check the recording consent and provider retention terms. Never use an AI-generated summary as the only record of an important customer commitment.

Your Chapter Exercise

Run and document eight past-behavior interviews.

1. Write the single decision the interviews will inform.
2. Recruit buyers who experienced the workflow recently.
3. Use the seven-section conversation guide.
4. Ask about the last occurrence, spend, workaround, consequence, and approval.
5. Record problem evidence separately from reaction to your solution.
6. Capture at least one boundary or trust requirement from each interview.
7. Write same-day notes with exact amounts or mark them unknown.
8. Label each buyer at evidence level 1 through 4.
9. Count qualified opportunities and test-ready next steps.
10. Decide whether to continue, change, or pause the investigation.

Chapter Deliverable

Keep an interview log and a one-page evidence brief containing:

• Number of buyers contacted and interviewed.
• Repeated trigger and consequence.
• Recent spend or workaround evidence.
• Decision maker and approval path.
• Privacy and quality boundaries.
• Evidence-level counts.
• Continue, change, or pause decision.

The next chapter turns a test-ready opportunity into a bounded offer and a request for a paid commitment.