Five linked hexagons: Empathize, Define, Ideate, Prototype, Test. Each has a one-line description. Empathize means learning about the person through observation, interviews, immersion and engagement. Testing is described as "another round of empathy work."

Empathy comes before Define, so the problem statement comes from evidence instead of being assumed. Testing feeds back into empathy, which makes the model a loop. Relay's interviews this week belong to the first stage, so they should not include a demo.

A photo of a paved loop around a round flower bed. A worn dirt path runs straight across the bed. The paved loop is labeled "what we think our customers need" and the dirt path "what our customers actually need."

The trail is evidence of behavior. No one was asked about it, and it records what people do when the designed route doesn't fit their goal. Look for Relay's equivalent: the sticky note, the personal cell phone or the side spreadsheet that front-desk staff use in place of the official waitlist.

An iceberg. The small tip above water is labeled EXPLICIT and the much larger mass below is labeled IMPLICIT.

People can state explicit needs on request, such as "a faster way to call the waitlist." Implicit needs, such as not looking incompetent to the therapist whose afternoon has a gap, usually have to be inferred from behavior and stories. The image claims they are larger. It doesn't measure them.

Concentric circles. The innermost ring holds the stated need, a book. The middle ring says "as a way to learn," and the outer ring says "as a way to discover her purpose." A tablet sits on the outer edge as an alternative way to meet the need.

Each step outward answers "why does that matter?" and admits more solutions. A tablet only appears as an option once you leave the innermost ring. Go up the ladder only as far as your evidence supports. Without episodes behind it, "discover her purpose" is speculation.

A large circle labeled "implicit needs lead to less obvious solutions" holds "a way to learn" and "a way to discover her purpose." A small circle at its edge labeled "explicit needs lead to obvious solutions" holds "ladder" and "book."

If you frame the request as the need, you get the ladder. If you frame the purpose as the need, you can get many responses. For Relay, "call the waitlist faster" leads to an auto-dialer. "Keep therapists' days full without front-desk heroics" also allows overbooking rules, standing short-notice lists and changes to the cancellation policy.

The question "who will benefit?" sits above an MRI-manufacturer example with three silhouettes: the procurement officer, the MRI mechanic and the patient.

The person who buys, the person who operates and the person who is affected often want different things. Relay has the same structure: a clinic owner who pays, a front-desk coordinator who works the waitlist and a patient who decides whether to take the slot. Interview each one separately.

Two photos. A pharmacy prescription bottle with a child-resistant cap sits on a counter, and then a serrated kitchen knife is used on the bottle on a cutting board.

The workaround is evidence that no interview would produce, because few people would volunteer that they attack their medication with a knife. Record what you see: the tool, the step and the setting. Only then interpret it, for example as limited grip strength, impatience or a design that blocks the intended adult user.

An OXO Good Grips vegetable peeler with its thick, soft, ribbed handle next to the brand name.

The widely told origin story is that OXO's founder designed the handle for a user with arthritis. The resulting grip was more comfortable for ordinary cooks too. The lesson is that extreme users show you a mechanism, here grip force. Market size still needs separate evidence. Relay's extreme users could be the clinic with the highest cancellation rate, or the coordinator who already fills nearly every gap by hand.

A list of interview rules: keep questions to about ten words, avoid yes/no questions, ask about specific occasions instead of what usually happens, ask why several times, avoid leading questions, listen most of the time, seek stories and allow silence. Examples include asking what someone ate for lunch today and following the reasons behind three hours of exercise.

Each rule protects the same thing: an account of real behavior that your expectations haven't shaped. "Do you usually struggle to fill cancellations?" invites a general opinion. "What happened with the last same-day cancellation?" gets an episode you can check.

Debrief prompts for interview pairs: what surprised you, which pain points and needs came up, what the person thought and felt, how the interviews changed your view of the product, and what one more question you would ask.

The last two prompts matter most. They make you record how the evidence challenged your assumptions and what is still unknown. Answer them the same day, before memory merges several interviews into one story.