How to Get Deeper Insights from Customer Interviews

‍‍Most teams walk away from customer interviews with opinions, not insights. This article explains how to fix that — by shifting from what users say they want to what their behavior actually reveals.

To bring an expert perspective into this, I spoke with Leslie Owensby, a product researcher with over 20 years of hands-on experience. Our conversation helped put words to a pattern I’ve seen come up again and again.

Teams talk to 20, 30, even 50 users and walk away feeling confident… only to realize months later that none of those conversations actually changed a single decision. The roadmap stayed the same. The assumptions remained the same. The product drifted in the same direction it was already heading.

So you default back to instinct. Or urgency. Or whatever decision was already half-made, forgetting that talking to customers doesn’t automatically produce insight.

Why Customer Interview Opinions Are Unreliable (And What to Use Instead)

First, start with a mindset shift. Insights are not opinions.

Most teams unconsciously interview for opinions. They ask what users think, what they like, and what they’d want in the future.

But opinions are slippery. People tend to be poor at predicting their own behavior, especially in hypothetical scenarios. When someone says “I would use this,” what they often mean is “I can imagine a version of myself who might.”

Real insight lives somewhere else entirely.

Probing Real Decisions and Contradictions

As a data analyst myself, I can validate that valuable insights can be found in historical data. The real value lies in 

• … behavior
• … past actions
• … moments where people had to choose one thing over another

That’s why questions like “Would you use this?” rarely help. They invite speculation. What changes everything is asking questions that pull people back into reality:

• “Tell me about the last time you tried to solve this.”
• “What did you actually do?”
• “What got in the way?”
• “What did you give up?”

When you ask that way, the conversation shifts. You stop collecting nice feedback and see how people really make decisions. 

And here’s something counterintuitive.

When a user’s stated values clash with their actions, you have found the signal. 

For example, a user might tell you they value "data security" above all else (Opinion), yet they currently share sensitive passwords over Slack (Behavior). The insight isn't that the user lied, but the discovery of a tension where convenience or workflow speed currently outweighs their intent to be secure. This "messy" honesty reveals where the real opportunity for a solution exists.

That’s what deeper insight feels like. It’s not cleaner. It’s messier but honest.

Breaking the Broken System Around User Conversations

Now, if this sounds obvious, here’s even a bitter pill to swallow. Even teams with good intentions still get interviews wrong.

• Founders lead questions without realizing it.
• They talk too much because silence feels awkward.
• They follow up deeply with one user and barely probe the next.
• They lose nuance while taking notes.
• They unconsciously look for confirmation.

And even when the conversation is great, nothing gets synthesized properly.

Transcripts live in one place. Notes live somewhere else. Key moments stay trapped in memory. Patterns never fully emerge. After interviewing Leslie, I found myself struggling with the exact pitfalls he described about messy transcripts and missed notes.

• Transcribing did not work 
• I recorded everything on the phone and revisited it 5 times to remember the details
• Forgot to take notes when Leslie mentioned something prominent

Moments like this push many teams to start believing interviews just don’t work — when in reality, the system around the interviews is broken.

The Prep Question That Changes Everything

Before any interview happens, there's one question that almost nobody asks, but everything depends on it.

What decision is this interview meant to inform?

As Leslie put it: "Good research should answer two questions: What do we need to know? And once we know it, what are we going to do differently?"

In short: if you can't define what you'd do differently based on what you hear, the interview won't produce actionable insight

The same goes for segmentation. Talking to “users” isn’t enough. Context shapes everything. Role, experience, alternatives, constraints. Without that, insights blur together and lose meaning.

During the interview itself, the hardest skill is restraint. Knowing when not to jump in. Knowing when to let someone struggle for words. Knowing when to follow the energy instead of the script.

And across interviews, consistency becomes critical. Insight comes from patterns and patterns only appear when questions are comparable. When results aren’t dependent on who happened to run the interview that day.

This is where even the strongest teams start to feel the strain.

The Hidden Costs of Traditional Interview Workflows

Weeks of hour-long interviews and days of preparation exhaust most teams, and Leslie admits that this operational friction eventually kills the research process entirely.

• Interviews are expensive in time.
• Interviews are hard to run continuously.
• Interview quality drops as volume increases.

Eventually, teams stop interviewing, not because they don’t believe in it, but because it no longer fits their workflow. The urgency of building takes over. Interviews become occasional, rushed, and reactive.

The operational friction is the actual bottleneck, not the lack of curiosity. 

And this is where using AI correctly can actually increase productivity.

Eliminating Bias and Bottlenecks with AI Interviewers

At some point, most teams hit the same walls.

• There’s no time to schedule interviews properly. 
• No dedicated research team  
• Agencies feel expensive and slow, especially when you’re trying to ship every two weeks.

That’s the gap AI interviewers are meant to fill.

Not by pretending to replace researchers, but by making enterprise-grade research behavior possible for teams that don’t have enterprise budgets, headcount, or timelines.

Beyond speed and budget, AI interviewers address a more fundamental issue: human bias. While no tool is a "silver bullet," AI offers a mechanistic advantage in three key areas:

• Standardized Probing: Unlike humans, who may inadvertently ask leading questions to avoid awkward silences or confirm their own hypotheses, AI follows standardized, neutral wording rules.

• Reduction of Interviewer Effects: Human participants often provide "socially desirable" answers to please a live interviewer. AI creates a lower-pressure environment that can encourage more honest admissions of failure or frustration.

• Auditability: Unlike a manual interview, where notes are filtered through a researcher's memory, AI allows for 100% traceability. You can audit the "logic chain" from a raw transcript to a final summary to ensure no insight is hallucinated.

In a nutshell, an AI interviewer doesn’t tell you what to build; it helps surface recurring themes and tensions in customer conversations faster, so teams can decide what matters with verification via transcripts and recordings.

From Surface Feedback to Real Product Direction

Customer interviews hold immense potential to transform product decisions, yet most teams emerge from dozens of conversations with little more than surface-level opinions. The path to deeper insights begins with a fundamental shift. Moving beyond what users say they want to uncover and the contradictions between their values and behaviors is the only way to uncover the customer needs.

Even with the best intentions, interviews often fail due to leading questions, inconsistent execution, confirmation bias, and fragmented synthesis.

When traditional methods become unsustainable, AI interviewers remove barriers of cost, availability, and human bias, enabling smaller teams to maintain enterprise-grade outcomes without slowing down.

Ultimately, deeper insights don’t come from talking to more users. They come from listening better, probing reality instead of speculation, and paying close attention to real human tension. When that tension is translated into clear product direction, customer interviews stop being just a research step and become one of your most reliable competitive advantages.