Opportunity discovery and market sizing · Lesson 3

Interview customers and map alternatives

6 min read

Lesson objectives

  • Interview customers and map alternatives: execute the named specialist task with a checked output.
Reading tools

Once a test segment is chosen, we need to understand the work customers are trying to accomplish. A discovery interview is neither a product pitch nor a vote on whether people like an idea. It is a structured way to gather accounts of the last time a problem appeared, what the customer did, what the current solution cost, and who was involved. “Would you buy this if we added a feature?” invites a polite imagined future. “When was the last cancellation, and what did reception do?” opens a sequence of behavior that can be checked. Strategyzer's experiment guidance distinguishes questions about past behavior from hypothetical opinion while also recognizing that interviews reveal what people say, weaker evidence than observed action.

Before the call, write one hypothesis and one learning decision. For example: “Late cancellations cost independent clinics material time. If a repeated pattern appears across clinics, test an automated reminder.” Do not turn the hypothesis into a script that prompts agreement. Prepare open questions: describe the last week when the schedule was crowded; how did you discover a cancellation; what did you do to fill the slot; who bears the cost of an empty slot? Probe for timing and quantities after the respondent answers, rather than supplying a number. A remembered estimate is not a measured count. Write “the manager recalled about five cancellations a week,” then seek an available, authorized record if the decision requires greater confidence.

Separate user and buyer roles. A receptionist may handle the schedule every day, while an owner approves software purchases and a technical lead checks integration before contracting. An interview with only the owner may reveal a budget but miss workflow failure. An interview with only the user may show the workflow but not the purchase condition. Put the speaker's role beside each statement; do not merge different voices into “the customer wants.” Do not collect patient details or images of sensitive records merely to enrich a business case. An abstract description or synthetic example is enough, subject to the organization's data-handling rules.

Code each note into three fields: job (“keep the practitioner's schedule usefully filled”), friction (“late cancellation or difficult reassignment”), and present alternative (“manual messages, shared spreadsheet, no action”). Add evidence strength. A particular event with approximate timing and a described response supports more than a general impression; an actual record or buying action supports more than one person's account. Strength does not judge a person's honesty. It asks whether the evidence can bear the cost of a decision. Preserve the person's words separately from your coding. The manager may have said, “I want fewer calls,” rather than “I want automated reminders.” The second statement is an interpretation of a possible solution and must not be attributed to the manager.

Compare alternatives by the customer's job, not by feature lists. A receptionist's messages may be flexible but consume after-hours time. An existing system may integrate with records yet be cumbersome to use. A clinic may do nothing because the cost of an empty slot is lower than the burden of change. Ask what happened the last time instead of buying another tool; the answer reveals the no-purchase alternative. In sizing the opportunity, do not turn everyone who is annoyed into a buyer. Look for a material consequence, a decision maker, capacity to change, and timing. These are discovery conditions, not a rigid universal qualification formula.

At the end of an interview, read back a short summary: “I understood that two late cancellations occurred last week and that you called three people on an informal list. Is that accurate?” A correction now costs less than developing an offer around a misunderstanding. After several interviews, actively seek a case that challenges your hypothesis. If three clinics report rare cancellations, they are not automatically “exceptions.” Your segment definition may be broad or your offer may be wrong. The specialist's decision at this stage is to run a small test, refine the segment, or pause the hypothesis. It is not a claim of product-market fit.

Watch how participants are recruited. Interview only people who complained before and the problem will look more common than it is. Interview only satisfied existing customers and you may miss the reasons noncustomers refuse to buy. Record invitations, replies, roles, and the characteristics you tried to vary, without pretending that ten interviews statistically represent a market. The purpose is to learn where the problem appears and where it does not. If a manager's account conflicts with an employee's, do not resolve it by rank. Ask about a shared event, inspect an authorized record if possible, and retain the disagreement if it remains. Each person may see a different part of the process.

Finish with an operational learning table: hypothesis, supporting evidence, contradicting evidence, a descriptive confidence level, next test, and deferred decision. For example, the claim that reminders reduce Clinic A's empty slots has a report of seven no-shows in its favor, but no measured count of slots refilled after a message. The next test could record empty slots before and after a defined reminder period while checking seasonal differences. A coincident improvement alone would not prove that the reminder caused it. This converts interviews from a collection of quotes into learning that can guide and check a bounded action.

Lesson summary

Interview customers and map alternatives with stated assumptions and a reviewable output.