How to Interview Doctors About Your Startup Idea

Interviewing doctors comes down to five moves: make the outreach ask an easy yes, frame the call as research instead of a pitch, run a tight 15-minute script, probe their actual workflow instead of their opinion of your idea, and debrief within the hour. Physicians will give you the truth in that window — but only if you don't waste it on a pitch they've learned to nod politely through.

Quick Answer: Treat a physician interview as the 15-minute slot it actually is, not a 45-minute coffee chat. Open with a specific, low-effort ask, spend most of the call on their last real case instead of your concept, and write up notes before your next call.

New to customer interviews generally? Start with the complete guide to customer research for pre-product founders, which covers the fundamentals this piece assumes. What follows is what changes when your interviewee wears a white coat, has a waiting room, and has spent a career being diplomatically noncommittal with people who aren't their patients.

Before You Book the Call: Prerequisites

Three things need to be settled before your first outreach message, or you'll waste the 15 minutes you eventually get.

Specialty targeting comes first, and it's narrower than founders expect. "Doctors" is not a customer segment — a hospitalist, a solo-practice dermatologist, and a hospital's director of population health experience your problem, and your budget, completely differently, even though all three hold an MD. Pick the exact specialty and practice setting your product targets, and interview only within that lane until a pattern emerges. Part of that targeting is understanding who actually pays for what you're building, since a hospitalist and a payer-side medical director sit on opposite sides of the same transaction — our breakdown of provider, patient, or payer economics in digital health is worth reading before you pick a specialty.

Credibility prep means one sentence, not a slide deck. You need a single, honest line explaining who you are and why you're asking — "I'm a founder researching how primary care handles prior authorizations" beats any pitch deck a physician will never open. A warm introduction from another clinician converts better than a cold message, but its absence just means your outreach in Step 1 has to work harder.

Scheduling norms differ from any other segment you've interviewed. A physician's calendar is chopped into fixed visit slots, not open blocks — realistic windows are before clinic starts, over a lunch that may not happen, or after the last patient. Expect a higher no-show rate than you're used to, and build slack into recruiting instead of treating each cancellation as a signal about your idea.

Step 1: Make the Ask an Easy Yes

The outreach message is a physician's first data point on whether talking to you wastes their time, so it can't read like a sales email. State the specific ask, time cost, and window up front: "I'm researching how [specialty] handles [narrow problem]. Could I ask about your last case like that — 15 minutes, whenever works, including evenings?" No attachment, no company description, no "pick your brain."

If you haven't identified who to send that message to yet, where to find doctors to interview covers ten sourcing channels ranked by cost and speed. This step assumes you already have a name and are converting them into a booked call.

Step 2: Frame It as Research, Not a Pitch

The first sixty seconds sets the register for the whole call, so the framing has to be explicit, not implied. Say, out loud, that you're doing research and not selling anything — physicians hear enough vendor pitches to assume you're one more unless you rule it out yourself. Ask permission before diving into specifics ("mind if I ask about the last time this came up?"), and if you want to record, ask that separately and be ready for a no. A physician who thinks they're being sold to gives you the polite version of the truth; one who thinks they're helping you understand a real problem tells you what actually happened.

Step 3: Run the 15-Minute Script

Structure the call so the time limit works for you, not against you:

• Minutes 0-2 — Frame and consent. State the research framing, confirm the time you have, and ask permission to dig into a specific recent instance.
• Minutes 2-11 — Workflow, in the past tense. Ask them to walk you through the last time the problem came up, start to finish. Stay here longer than feels natural — this is where the real evidence lives.
• Minutes 11-13 — One reaction question, asked last. Only once you understand what happened, ask the single question closest to your concept, framed against their specific case.
• Minutes 13-15 — Close and referral ask. Thank them concretely, ask if they know one other person in a similar role, and confirm whether a follow-up is welcome.

Order matters more than any single question. Ask about your product before their workflow, and every later answer gets filtered through "what would make this founder happy" — the exact bias you're trying to avoid.

Step 4: Probe Their Actual Workflow, Not Their Opinion of Yours

This is the Mom Test rule in clinical form: ask about specifics in their past, not opinions about your future product. Rob Fitzpatrick's The Mom Test argues that most people are inclined to be diplomatically encouraging about an idea they're asked to react to — and physicians, trained for years to be measured and reassuring with patients and colleagues alike, are exactly the population that rule was built for. "Would this be useful to you?" gets a polite yes from nearly anyone. "Walk me through the last time this happened" gets a fact.

Push past the first answer. When they describe a workflow, follow with what it actually cost — time delayed, skipped, or pushed after hours — and whether it's a one-off or something that happens most weeks. A single anecdote is a data point; a pattern only shows up once you ask "has this happened more than once this month" and get a real answer.

Step 5: Debrief Within the Hour

Write up raw notes before your next email, not at day's end once three conversations have blurred together. Separate what the physician actually said from your interpretation — quotes in one column, your read in another — and capture specific numbers and phrases verbatim rather than a cleaned-up paraphrase. Flag anything that contradicted your working hypothesis explicitly, in writing, before the urge to smooth it over sets in. Debriefing immediately is what keeps interview six from quietly reshaping your memory of interview one.

Questions That Work vs. Questions That Trigger Polite Lies

The table below pairs common founder questions with a Mom Test-style rewrite. Every question in the left column can be answered with a comfortable "sure, that sounds useful" — a physician's default register with anyone who isn't a patient or colleague — while the right column forces a specific, checkable answer.

The pattern on the right is the same every time: past tense, specific instance, checkable detail. If a question can be answered honestly with a shrug, rewrite it until it can't.

A Worked Example: An Annotated Clinician Interview

The exchange below is a composite, illustrative example — not a transcript of a real interview — showing how the Step 3 script plays out for a founder researching how primary care physicians handle prior authorizations.

Founder: "Thanks for making time between patients — I know you've got maybe fifteen minutes. I'm researching how primary care handles prior authorizations, not selling anything today. Mind if I ask about the last one you dealt with?" Names the time constraint, states the non-pitch frame, and asks permission — all before question one.

Physician: "Sure, go ahead."

Founder: "Walk me through the last prior auth you personally had to deal with, start to finish." A single past-tense, specific-instance question — the Step 4 move. Not "how do you feel about prior auths in general."

Physician: "Monday. A patient needed an MRI, the insurer wanted six weeks of documented failed conservative treatment first. My MA started the fax, it bounced twice for a missing code, so I finished it myself between patients."

Founder: "When you did it yourself, what did that push out of your day?" Digs for the real cost — a delayed lunch, patients seen later — instead of accepting "it was annoying" as a complete answer.

Physician: "I ate lunch at four and charted two more patients that night."

Founder: "Has that happened more than once this month?" A frequency check, still past tense, turning one anecdote into either a pattern or a one-off.

Founder (closing, minute 13): "One more thing — if something had caught that missing code before the fax bounced, would that have changed how Monday went?" The one product-reaction question in the whole call, asked last and anchored to the specific case just described.

The useful evidence afterward is the fax bouncing twice and the four o'clock lunch — not the "yes, that would have helped." The workflow detail is specific and checkable; the reaction to a hypothetical fix is neither, and gets logged as a lead worth following up, not as validation.

Mistakes That Burn Clinician Goodwill

Physician networks are small and word travels inside them, so a bad interview costs you more than one contact.

• Running past the 15 minutes you asked for. A patient is almost always waiting; going over reads as disrespect for the constraint you were told about, not enthusiasm.
• Pitching before asking a single workflow question. If your first substantive question is about your product, you've told them this was a sales call in disguise.
• Recording without asking first, every time. Never assume consent carries over from a prior conversation — ask explicitly before you hit record.
• Using jargon that signals you don't understand clinical practice. It doesn't read as sophisticated; it reads as skipped homework.
• Asking for a referral to a named colleague with no easy way to decline. Ask if they know "someone in a similar role," not "can you introduce me to Dr. [Name]."
• Following up aggressively after a decline or silence. One polite follow-up is normal; a second and third read as exactly the pressure you claimed you weren't applying.

Key Takeaways

• Doctors give you 15 real minutes, not 45 relaxed ones — script the call to fit that window.
• Frame the call as research, explicitly, in the first sixty seconds, or physicians default to treating it as a pitch.
• Ask about a specific past instance, not a future opinion — the Mom Test principle applies with extra force to a profession trained to be diplomatically encouraging.
• Save your one product-reaction question for the last two minutes, anchored to the case they just described.
• Debrief within the hour, separating what was said from what you inferred, before the next call blurs the details.
• A bad interview costs more than one contact in a small specialty network — spend physicians' time and goodwill deliberately.

Frequently Asked Questions

How long should a physician interview actually be?

Plan for 15 minutes and treat more as a bonus, not the baseline. Physicians schedule around fixed visit slots, so a 15-minute ask is far easier to say yes to than an open-ended "got a few minutes to chat" — and the Step 3 script fits real evidence into that window.

Do I need a warm introduction to get a doctor to talk to me?

It helps but isn't required — a warm intro converts better than cold outreach, but plenty of physicians say yes to a specific, low-effort, well-framed cold ask. If you're still building a list of people to reach, where to find doctors to interview covers channels that don't depend on already knowing someone.

Should I pay physicians for a 15-minute interview?

Many physicians will do a short, informal conversation for free, especially with a clear research frame and a respected time limit — but paying market-rate consulting fees removes friction and signals you value their time, particularly for longer sessions or scarcer specialties. Recruiting through a paid expert network usually bundles payment into the channel already.

How many doctor interviews do I need before I trust a pattern?

There's no universal number — what matters is whether you're still hearing new workflow details or the same ones repeating. A practical signal: once several consecutive interviews within the same specialty and setting stop surfacing anything new, you're likely looking at a real pattern rather than a handful of outliers.