When to Sit in on Discovery, Not Just the Demo

The CEO of a Series B startup I worked with had a standing rule: she attended every sprint demo. She was good at it — sharp questions, clear feedback, occasionally a “why didn’t we do X instead?” that sent the team scrambling. The problem wasn’t her presence at demos. The problem was her absence everywhere else. She didn’t sit in on customer interviews. She didn’t attend discovery sessions. She didn’t read the opportunity solution trees the PM had been building for two quarters. So when she showed up at demo time and asked “why didn’t we do X instead?”, the answer was always the same: “because we learned in discovery that customers don’t actually want X.” And she’d nod, and accept it, and the next sprint she’d ask the same question again, because she’d never been in the room where the learning happened. Her mental model of the customer was frozen at the Series A pitch deck, and demo-time feedback was her only mechanism for updating it. The cost was invisible but real: the team spent fifteen minutes per demo translating customer context to someone who’d missed three months of it, and every “why didn’t we do X?” was a decision revisited that had already been resolved.

The fix wasn’t complicated. She started attending one discovery interview per month — not every one, not even most, just one. The effect was disproportionate. Her demo questions changed from “why didn’t we do X?” to “I noticed in the interview that customers struggle with Y — are we addressing that?” The translation overhead at demos dropped to near zero because she’d heard the customer’s words herself. And the team stopped dreading demo day because the CEO’s feedback was grounded in shared context instead of a parallel reality.

The problem with demo-only leadership

Demo time is the most expensive time to learn about your customers. By demo day, the sprint is done, the code is shipped, and the cost of changing direction is the full price of rework. Discovery interviews happen three months before demo day, when the cost of changing direction is the price of a conversation. The gap between those two moments is where most product waste lives: not in bad execution, but in correct execution of the wrong thing, discovered too late to cheaply correct.

Founders and execs who only see demos are operating with a lagging indicator of customer reality. They see what the team built based on what the team learned — a filtered, interpreted, two-month-old version of customer needs. The filter isn’t dishonest; it’s structural. The PM synthesizes interviews into opportunities, the opportunity tree shapes the roadmap, the sprint picks from the roadmap, and the demo presents the sprint output. Each layer compresses and contextualizes, and by the time it reaches the exec, the raw customer voice is gone. What remains is the team’s interpretation, which is usually right but occasionally wrong in ways that only the original customer words would reveal.

The specific failure mode I’ve watched most often: the exec’s mental model drifts from the customer’s actual reality, and the demo is the only feedback loop. Drift is gradual — six months of not hearing customers means six months of decisions made against an outdated model — and the demo can’t catch it because the demo is built on the same outdated model. The exec reviews work that was shaped by their own stale assumptions, sees no dissonance, and approves it. The customer discovers the dissonance three months later when the feature ships and nobody uses it.

The right moments to attend

Not every discovery interview needs an exec in the room. Presence changes the dynamic — the customer speaks differently when the CEO is listening, and the PM’s facilitation style shifts when they’re performing for an audience instead of exploring a problem. The goal is selective presence at moments where the exec’s update is most valuable and the disruption is most contained.

The first interview with a new segment. When the team enters a new customer type — a new industry, a new company size, a new use case — the exec’s mental model is almost certainly wrong. They have beliefs about this segment from sales conversations, board decks, or market research, and those beliefs need to be tested against real voices. Sitting in on the first two or three interviews with the new segment gives the exec a direct signal instead of a PM’s summary. The PM runs the interview; the exec observes and asks follow-up questions at the end, if the customer is comfortable.

Interviews that surface a strategic pivot signal. When a customer says something that challenges the company’s direction — “we’d pay twice as much for the opposite of what you’re building” — the exec needs to hear it raw, not translated. The PM will flag it in the opportunity tree, but the flag is an interpretation. The exec hearing “we’d pay twice as much for the opposite” directly updates their model in a way that a Slack message about “customer preference shift” doesn’t.

Interviews where the customer’s language differs from the company’s. Companies develop internal vocabulary that diverges from customer language over time. When the team calls something a “workspace” and the customer calls it a “project board,” that’s a positioning gap. The exec hearing the customer’s language directly — not through a PM’s glossary — catches divergence that the team has normalized.

Quarterly strategy interviews. Once per quarter, the exec sits in on a discovery interview that specifically tests the strategic direction: “are we still solving the right problem for the right people?” This isn’t a product interview; it’s a strategy-validation interview. The exec’s presence signals to the team that strategy is grounded in customer reality, not board decks.

The wrong moments to attend

The same presence that’s valuable in some contexts is destructive in others.

When the exec can’t resist pitching. Some founders hear a customer problem and immediately start selling the solution. “Oh, we’re building exactly that — let me tell you about our roadmap.” This kills the interview: the customer shifts from describing their reality to reacting to the exec’s pitch, and the PM loses the discovery signal. If the exec can’t resist this, they’re not ready for discovery attendance. The fix is a pre-interview agreement: “you’re here to listen, not to present. Questions at the end only.”

When the power dynamic changes the customer’s honesty. A customer talking to a PM is peer-to-peer — the customer feels safe saying “your product is confusing.” A customer talking to the CEO feels like they’re talking to the brand, and they soften their criticism. “It’s fine, I just wish it had feature X” instead of “it’s confusing and I almost churned.” If the interview is about honest friction, the exec’s presence may reduce honesty. The PM should flag this in advance: “this customer might be more candid without an exec in the room.”

When the exec turns it into a steering meeting. Discovery interviews are for learning, not directing. When the exec starts telling the PM what to build based on the interview — “we should definitely do the dashboard thing” — the interview becomes a mini-steering meeting and the PM’s discovery discipline breaks down. The exec needs to understand that the interview is a data-collection exercise, and the synthesis happens later, separately, in the opportunity tree.

The mechanism that makes it work

The exec attending one interview per month works only if three things are in place.

A shared read-ahead. Before the interview, the PM sends the exec a one-page brief: who the customer is, what we’re exploring, what we already know, what we’re trying to learn. This takes five minutes to write and saves fifteen minutes of context-setting in the room. The exec walks in informed, not cold.

A post-interview debrief. Within twenty-four hours of the interview, the PM and exec spend ten minutes discussing what they heard. The PM shares their synthesis; the exec shares what surprised them. This is where the mental-model update actually happens — not during the interview, but in the conversation about it. The exec’s surprises are signal: they mark where the company’s beliefs diverge from customer reality.

A standing agenda item at demo day. “What we learned from customers this sprint” becomes the first slide of every demo. The PM shares verbatim quotes, opportunity-tree updates, and shifts in understanding. The exec, having attended at least one interview, has a reference point for evaluating the summary. The demo shifts from “here’s what we built” to “here’s what we learned and what we built because of it.”

The compound effect

The founder who attended one interview per month didn’t become a product manager. She didn’t start writing stories or prioritizing the backlog. What changed was the quality of her decisions at the strategic level. Her board questions shifted from “are we on track for revenue?” to “are we solving the right problem?” Her hiring decisions reflected current customer needs instead of last year’s assumptions. Her product reviews started with “what did we learn?” instead of “what did we ship?”

The compound effect is trust. When the team knows the CEO has heard the customer directly, they don’t need to translate customer context in every presentation. When the CEO trusts the team’s discovery process because they’ve witnessed it, the approval cycle shortens. When both sides share the same customer vocabulary, the “why didn’t we do X?” questions disappear — not because the CEO stopped asking, but because they already knew the answer.

The investment is one interview per month and a ten-minute debrief. The return is a leadership team whose decisions are grounded in customer reality instead of a two-month-old interpretation of it. That’s the highest-leverage hour a founder can spend on product quality, and most founders never discover it because they’re too busy attending demos.