Running Customer Discovery Interviews

Imagine you are leading a customer engagement for the first time. A more experienced colleague sits beside you on the calls, and you are responsible for the questions, the schedule, and the result. Halfway through the discovery work, that colleague gives you feedback that cuts to the heart of the job: you were asking the questions you had already written down, not the questions the customer's own words were pointing toward.

The account behind this moment is a real first engagement that we will follow through the rest of this article. It was a three-week effort for a mid-market insurance carrier, building a single workflow that summarizes claims for a defined class of claims. Along the way the lead learned three lessons that apply well beyond that one project: how leadership is handed over gradually, how to listen well enough to change course mid-interview, and how a schedule can be lost in places nobody planned to look.

Keep these three threads in mind as you read: who holds leadership at each stage, what the lead does when a customer answer points somewhere the plan did not, and where the schedule actually slips.

The lead's role is the one that carries the responsibility for the work. In this account the lead was responsible for the questions, the schedule, and the result. The senior shadow's role is different. He observes closely enough that his later feedback can be specific and useful, and when the moment matters he offers a correction and then helps the lead act on it. Mentorship is the bridge between observing and acting.

Before reading the answer, think about the problem the lead faced. In the second interview she had prepared a list of questions and worked through it. The senior shadow, watching without speaking, later pointed out what he had seen.

The lead did not stop preparing. She still walked into each conversation with a plan. What changed was her willingness to set the plan aside when a customer's answer opened a more useful thread. The account describes her practicing this one habit through the rest of the engagement, which is how a single piece of feedback becomes a durable change.

The flow below is an illustrative model of this decision. It is not a procedure taken from the source. At each answer, the interviewer asks whether the answer points toward a useful follow-up. If it does, the next question is built from that answer. If it does not, the interviewer returns to the next topic on the prepared list. In both branches, the interviewer then listens to the next answer and repeats the check.

Defining the metric is only the first step. The team also needs a threshold, the level the metric must reach before the customer will accept the result. Here the threshold was thirty percent. During the first week of validation the metric reached twenty-three percent, and the customer then asked for the remaining seven percent.

Schedules in customer engagements rarely fail at the point where the work is hardest. The first slice reached the customer's staging account on day forty-eight, two days after the planned date, because of a data-permissions delay. That delay was real, but it was also solvable. The lead resolved it by escalating through her sponsor, the internal person with authority to unblock it, with Theo's help.

The later and larger delay came from the revision cycle. The model and prompt changes needed to close the gap to the threshold took about three days of actual engineering work. The validation extension, however, lasted two weeks. Most of that time went to getting revised work onto the customer's staging environment, not to the revisions themselves.