I played them seven minutes of a film. The room went silent. What happened next told me more about their gaps than a year of design reviews.
The scene is from Moneyball. A baseball team's management meeting — loud, chaotic, several people talking across each other, a room full of competing agendas and unspoken fears. I asked my team to watch it. Seven minutes and fourteen seconds. No phones. Just watch.
The room went completely silent. Not out of politeness — they knew questions were coming. You could feel them watching carefully, storing details, preparing.
One person on the team has a hearing impairment. Midway through, we realised the audio wasn't reaching them the way it needed to. We paused, searched for a Chrome extension, got captions running. Nobody complained. We figured it out in the room, together, and kept going.
I mention this because it was the first small lesson of the session, before any question had been asked: a room that can't hear the conversation properly cannot read it accurately. Accessibility and observation are the same problem.
The First Reaction
The moment the clip ended, the energy shifted. Eager — almost impatient to answer. They'd been holding their observations and wanted to put them somewhere.
Then I gave them the questions. Some answerable directly from what was said. Others requiring inference — reading between the lines, connecting tone to meaning, understanding what each person in that room was actually protecting.
One person, before they'd even looked at the questions properly, said: "There's no right answer to these."
I know this person well. They have strong opinions — genuinely strong, well-considered opinions. But they have a habit of softening them in group settings. Of talking around the edges of what they think rather than landing on it directly. Of saying things like "I might be wrong but..." before saying something that is, in fact, correct.
"There's no right answer" was that same instinct. A way of protecting themselves before committing to a position. Because if there's no right answer, you can't be wrong.
I pushed back gently: there are limited right answers. Not infinite options. Not anything goes. The questions weren't asking for feelings — they were asking for reading. And reading can be accurate or inaccurate.
They had already answered. All of them had. The questions were multiple choice. The exercise was complete.
What they meant was: they couldn't yet see how the answers were possible. They had watched the same seven minutes. But the connection between what was said out loud and what it meant — about the people saying it, the dynamics in the room, what was actually being decided — that connection hadn't fully formed.
They were transcribing. Not translating.
One Character, Three Readings
The sharpest moment in the debrief came from a debate about a single character. The scene's central figure — a decision-maker pushing a new direction in a resistant room — produced three completely different readings from my team.
One said: confident.
Another said: rigid.
A third said: overconfident, not listening to others.
I watched a couple of the quieter people in the room shift visibly as the debate unfolded. You could see them recalibrating. Not uncomfortable — but unsettled in the productive way that happens when you realise your first reading might have been incomplete. Not "am I wrong?" exactly. More like: "wait, am I seeing the same thing they're seeing?"
My own reading was different from all three. The character was the decision-maker in the room. Everyone else was someone they needed to bring along. Brief, controlled directness in that position isn't rigidity or arrogance — it's the specific behaviour of someone who knows they're about to make an unpopular call and has stopped pretending the conversation is going somewhere useful. The room was uncomfortable. People were speaking in one or two words, not because of the decision-maker's overconfidence, but because of their own uncertainty about what was being asked of them.
The confusion my team felt wasn't about the character. It was about power. What does authority look like when it's being exercised, not performed? How do you read someone who isn't trying to make you like them?
These are questions that come up in every stakeholder meeting, every client review, every room where someone senior is making a decision that not everyone agrees with. Learning to read them accurately — not emotionally, not through the lens of your own relationship with authority — is a design skill.
The Transcript Exercise: Who Did They Misread?
The second activity used a written transcript — a fictional product meeting about a B2B analytics platform. The dashboard had technically been adopted. Feature usage metrics looked fine. But customers were exporting raw CSVs and building their own reports in Excel. Renewal conversations were getting harder.
I asked my team to read the transcript and identify what was actually broken. Not the stated problem — the real one.
The characters they misread most were the PM and the senior leadership. The Customer Success Managers were the trickiest — my team was still discovering what the CSM role even means, what their relationship to the product is, whose interests they represent. Without that context, they couldn't accurately read what the CSMs were doing in the conversation or why it mattered.
The most interesting misread: the leader in the transcript asks short, pointed questions throughout. My team mapped this directly onto leaders they'd encountered in their own company. They read the behaviour through a familiar lens rather than through the logic of the scenario.
I had to name that explicitly: the way you've experienced a certain kind of leadership in your specific context is not a universal template. When you're in a room with a client, a stakeholder, a senior figure you haven't met before — your job is to read them as they are, not through the filter of someone else they remind you of.
What Happens in Real Rooms
After the workshop, I sat with something that had been bothering me for a while.
I've been in client calls and stakeholder reviews where I've watched my team miss the subtext in real time. Someone asks a question with visible frustration underneath it, and my team answers the surface question — giving a basic, technically correct response — because they want to contribute something and can't locate where the real friction is.
I've seen them miss the whole context. Not because they weren't listening, but because they didn't have the confidence to pause the conversation and ask for clarification. To say: I want to make sure I understand what you're asking before I answer. Instead, they stay in motion — answering, proposing, responding — and an hour later, after the meeting ends, they come to me privately and ask what a particular phrase meant. A phrase that was, in most cases, the foundation the entire conversation had been built on.
An hour in a meeting where you've missed the context isn't just unproductive. It's a signal to the room that you're not tracking.
But the thing that has stayed with me most is something subtler. In meetings, several of my team members look at me before they ask a question. They make eye contact — checking, waiting for a small nod. If I give it, they proceed with confidence. If I don't, they pull back.
This tells me something important: their confidence in their own reading of the room is conditional. They don't yet fully trust what they're observing. They need external confirmation before they act on it.
That's what I'm trying to change. Not by telling them to be more confident — that doesn't work. By building the skill that makes confidence earned: the ability to observe accurately, to distinguish between transcription and translation, to know the difference between what was said and what was meant.
Words Lie. Behaviour Doesn't.
The frame I use for all of this is simple: words lie, behaviour doesn't.
What someone says in a meeting is often a performance — of confidence, of alignment, of enthusiasm they may or may not feel. What their body does is the real data. The pause before they agree. The question asked twice in slightly different words because the first answer didn't land. The way someone's posture changes when a specific topic is raised.
The three signals I ask designers to watch for: hesitation, backtracking, over-scanning. Hesitation usually means an unclear workflow. Backtracking means a broken mental model. Over-scanning means too much information competing for attention. These don't appear in surveys or feature requests. They live in the seconds between actions.
The same signals exist in meetings. In client calls. In the moment a senior stakeholder asks a question that sounds simple but isn't. Learning to read them — in real time, while also participating in the conversation — is the skill that separates a designer who gets called into strategic discussions from one who only gets called in to present work that's already been decided.
What Observation Is Actually For
I built the observation exercise because I kept seeing the same pattern: talented designers with good instincts, struggling to act on those instincts in high-stakes rooms. Not because the instinct was wrong. Because they hadn't yet built the confidence that comes from knowing how to read a room systematically rather than just feeling it.
Observation is not a personality trait. It is a skill. It can be practised with a film clip, a transcript, a set of questions that force you to distinguish between what you saw and what you assumed.
And when it's developed — when someone on your team stops looking at you for a nod before they ask a question, because they've learned to trust their own reading of the room — that's the shift that changes everything about how they work.
Observation tells you what's broken. But knowing it isn't enough. You still have to make a room full of people feel it too. That's a different skill entirely — and it starts with learning to tell a story that lands.