You just wrapped a customer interview. You’ve got two pages of notes, and when you read them back later that day, nothing jumps out. No surprise. Nothing you didn’t already believe walking in.
That’s not really an empathy interview. It’s more like a survey, just done on Zoom.
I’m Kenny Kranseler, a principal consultant and trainer here at Productside. Before this, I spent years doing product management at places like Kentucky Fried Chicken, Microsoft, Amazon, and a string of startups around Seattle. I’ve run more empathy interviews than I can count, and I’ve watched just as many go sideways. Recently I ran a live workshop called Ask, Listen, Learn: Running Better Empathy Interviews with AI, and one of the first things I asked the room was how many empathy interviews they’d personally run in the last quarter. About 40% said none. If that’s you too, you’re not alone, and it’s exactly why I want to walk through what really works.
Why Empathy Interviews Still Matter
There’s a quote attributed to Henry Ford, probably 120 years old at this point: “If I had asked people what they wanted, they would have asked for a faster horse.” If we only asked customers what they wanted, we’d still be waiting on faster horses instead of driving cars.
Most PMs default to asking “what do you want,” which surfaces feature requests, not needs. And if you’re getting feedback secondhand, through customer success or sales, they’ve already put a filter on it before you ever see it. Empathy interviews get you past both problems. You hear the functional, social, and emotional jobs your customer is actually trying to get done, not just the feature they think will fix it. Maybe they want to finish a task faster. Maybe they want to walk into a meeting feeling confident instead of hedging. Maybe they just want to look like they’re on top of things. None of that shows up in a feature request.
The Trap: Two Ways to Lose the Story
There are two ways product teams lose the plot on empathy interviews, and they sit at opposite ends of the spectrum.
- One is handing the whole thing to AI: synthetic personas, AI-run interviews, zero human involvement. It’s fast, cheap, and always available. It’s also hollow. You lose the real story, and you can never be fully confident you’re hearing from the right people.
- The other is over-preparing: a rigid, scripted list of questions asked the same way every time. That does just as much damage from the opposite direction. It boxes out the customer’s story before it has a chance to unfold.
The fix sits in the middle: a loose topic guide instead of a script.
Where to Start
Before you talk to anyone, get specific about who you’re talking to. In the workshop, I took the role as a product manager for a fictitious CRM product and created persona named Priya, a sales operations manager who owns a weekly pipeline forecast across 40 reps. Functionally, she needs an accurate forecast built from rep input. Emotionally, she wants to walk into her VP’s office without hedging and be seen as the source of truth on revenue data.
I gave AI her bio, her pains, her goals, and asked for four or five loose topic areas to explore, not specific questions. That gave me a topic guide covering how forecasts get built today, what “trustworthy data” means to her, moments where the numbers went wrong, what happens when leadership pushes back, and what she wishes she didn’t have to double-check.
Grab the exact persona brief and prompts I used. Download the free prompt pack here.
How to Conduct an Empathy Interview
Practice first. Don’t jump straight into the real conversation.
Make sure you start with open-ended questions; closed questions get you a yes or no and nothing else. “Why” is your best friend here, and so is “tell me more” when silence needs filling. Speaking of silence: when you feel the urge to fill a quiet moment, don’t. People have a natural instinct to fill silence themselves, and that’s usually when the best stories come out. I’ve had colleagues tell interviewees upfront, “I’m going to ask questions you feel like I should already know the answer to, but I’m here to learn from you.” That one line buys you permission to ask the obvious stuff.
During the workshop, I opened Claude live and used it to walk through a empathy interview role-play as Priya, then had Claude critique my interviewing performance, flagging every question I asked that was leading, closed, or jumped straight to a solution. That’s where AI is valuable in conducting the empathy interview: practicing the conversation before it counts, not having the conversation for you.
Should I interview customers one-on-one or chat with a group of them? Both can work, but know what you’re trading off. A group can bounce perspectives off each other, but you risk getting groupthink, as people consolidate around one loud opinion. Watch for power dynamics too. I’ve seen a user clam up completely because their boss was sitting in the same interview. If you need candor, go one-on-one.
When Stakeholders Push Back on What You Heard
Here’s a scenario that came up in our live Q&A during the webinar: a PM attendee ran an interview facilitated by a customer success rep, shared their takeaways with a VP, and the customer success person disagreed, causing a conflict where the VP sided with the customer success rep who’d “been here 20 years” over the newer product person.
Two things can help prevent this:
- Transcribe every interview; it gives you documented evidence instead of a he-said-she-said argument, and it becomes fodder for consolidated insights across all your interviews later.
- And debrief immediately, with anyone who joined you, before either of you drifts toward your own interpretation.
Be careful who you bring into the room in the first place, too. Salespeople are sometimes your only path to a customer, but selling and empathy are different skills. I’ve had to tell a sales team flat out: I need thirty minutes alone with this customer, or I’m not coming.
How AI Interview Analysis Helps
Once you’ve got a series of empathy interview transcripts, AI earns its keep. Feed it even a single interview and ask it to summarize stated needs, underlying goals, and pains and gains, with the specific quotes behind each one. AI interview analysis turns hours of note-sorting into minutes, especially once you’re working across a dozen transcripts instead of one. Getting a specific supporting quote used to mean digging through pages of handwritten notes. Now you can just ask for it.
Take it further: set up a project (Copilot calls them agents, Claude and GPT call them projects, Gemini calls them gems) loaded with all your interview transcripts and debrief notes. Let stakeholders query it directly: “Show me a conversation where a customer felt this way.” That gets your whole team closer to the customer, not just the person who ran the interview.
Build AI Habits That Compound
A few habits separate product managers who get real value from AI here from people who get generic noise back. Context matters most. Claude didn’t do anything useful for me until I’d fed it Priya’s bio, her goals, and what I was trying to learn. Without that, you’re asking a stranger to guess.
Treat it as a teammate you’re briefing, not an oracle you’re consulting. It only knows what you tell it. Structure your prompts consistently so the output comes back in a form you can use session over session. And don’t ask for a pile of random ideas; ask it to work toward a specific outcome. None of this is a one-time setup. You’ll get better results the more you iterate and correct it.
The One Thing AI Can’t Do For You
This is where I ended the workshop, and it still holds: AI will coach you and synthesize what you hear. It will not have the conversation for you. Don’t build a roster of synthetic customers and call it research. Go talk to real people, in person or live on a call, and let AI sharpen your empathy interviews skills instead of standing in for them.
- If you want the full session (including the live Claude role-play with Priya, the audience Q&A on handling stakeholder pushback, and the exact prompts used throughout) watch the on-demand webinar Ask, Listen, Learn. It walks through the complete framework for using AI to prep, conduct, and analyze empathy interviews.
- If you want to go further (not just running better interviews, but building the discovery habits that compound across your whole product career) you’ll want to join our AI Product Management course.
- What’s the AI habit you’ve built into your own customer interviews, or the one you’re still avoiding? Share it on LinkedIn and tag @Productside. We’d love to hear how you’re using AI to listen better, not just faster.
