Being the "wrongest" in the room

Transcript

‍Katie: Rhys, welcome to Unite Voices. We're excited to have you on the show. Give us a little intro — who are you?

Rhys: Thanks for having me. It's nice to be on a platform that brings science and empathy back into the fold — that's what I'm here to do, and you have a very similar mission.

Short version: I graduated into a recession many years ago. We thought we had it bad — but the next generation of marketers coming out is going to have it rough with uncertainty, with barriers changing every week. So I've invested in helping the next generation figure this out: find the evidence, rely on their empathy to really listen to customers and each other, and grow their business with a slightly different model. That's the plan.

I'll also call out — I'm wearing my NASA shirt, celebrating the Artemis mission that just made it safely back from around the moon. Big science nerd, my dad was a science teacher, I come by it naturally. A couple of firsts worth noting: first woman breaking the glass ceiling and smashing through lower orbit — I'm a dad of two girls, so we've been watching closely. And first Canadian out of low orbit. Captain Jeremy Hansen going around the moon, that's amazing for us. I grew up knowing Chris Hadfield was the most famous Canadian, then Ryan Gosling came along — and now Hansen. I needed that story right now. It felt like a booster rocket for my soul in the current news cycle.

Katie: I love it. And it's actually relevant to experimentation. I went to a conference last week and one of the speakers said: "I don't want to be called an innovator, because innovators claim they know everything. I want to be an explorer." That ties right back into your philosophy. We're in an era where everyone wants to be right — but being wrong is a huge competitive advantage. Tell us about that.

Being the "wrongest in the room"

Rhys: "Being the wrongest in the room" is provocatively titled, but it's the counter to trying to be the smartest in the room — which isn't what we're about anymore. And leaders are starting to recognize that.

It wasn't a title I made up, by the way. I was given it by the head of Canada at a company I worked at. On my third day, I walked in and he said, "You're the wrongest in the room." I froze. He said, "Relax, it's a compliment. You were asking all these questions, you were willing to look green, you were willing to look wrong and figure it out — we need more of that."

I've really embodied that. And it takes some explaining, especially for leaders trying to organize their teams in this new AI era. Being wrong in the room means going first. It means being the leader who shares a raw idea, goes to the whiteboard, and says, "Hey, I've been thinking about this — what do we think?" For teams that aren't practiced at that yet, it's so important that a leader sets the table first.

Here's the case:

1. Psychological safety and being wrong go hand in hand. It's giving other people in the room the space to share their ideas — because that's where the ideas are. The two-to-three-person growth team is gone. We're doing high participation now. Everyone has a say.

2. Creative ideas give you stronger signals to begin with. You can have all the statistical rigor in the world, but you can't apply a statistical model and make a bad idea win. You need great, team-generated ideas to get a signal in the first place. We're trying to automate too much. We're pushing things off to AI in places where we should be leaning into thinking.

3. Stronger signals resolve faster. You can measure faster, get bigger results, move faster — that's a competitive edge.

4. You argue less. If you're stuck looking at the same data set and you're not sure how to interpret it, experimentation is your tool. You can say, "Look, we're stuck. Let's run a test. We'll reconvene in a week and the market will help us make our minds up."

Big tech brands talk about hiring the best talent, moving fast, being agile. The idea that experimentation is supposed to be so difficult acts as a moat for them. It's not. Every team can run experiments. The model looks different at a startup versus an enterprise — of course it does. But it's still experimentation. There's still rigor. And you still set the table for your team to take guesses and be willing to be wrong with new evidence.

From decider to facilitator

Katie: I see a lot of authority and gravitas that comes with being a decision-maker. But how do we shift from being decision-focused to being more of a facilitator? Because that's what I'm hearing from you.

Rhys: As teams approach mid-size — funded, founders' hands in everything — you have to move from founder-led decision-making to team-led, autonomous decision-making with some structure. Structured autonomy is where the biggest teams move the fastest.

Leaders have to go first — like in any culture change. Bring your raw ideas to the whiteboard. I call it "to the boards" — hockey reference — from the whiteboards to the boardroom. Share your stuff and show that it's safe to share a raw idea.

A few specific shifts:

Get rid of idea ownership. How often do we say, "Hey, what was Ted's experiment we were going to run last week?" Attaching Ted to the idea — even if it was his — makes it difficult for the rest of the team to see themselves in that experiment. It puts pressure on Ted. He gets defensive because now he needs his test to win. That's not how we want to look at innovation. With the right environment, the right creative rituals, the right facilitation, your team will eventually come up with that idea anyway. Ted got there first — but you can remove the ownership.

Debate the premise, not the person. "Katie, you're wrong" — ouch. We don't want to hear we're wrong. But it's not us — it's our idea that may be debated. Instead, say: "I don't think I agree with that premise. I don't agree that more users are doing this, or that revenue is an opportunity over here." Now we're on the same side of the table, debating the premise together.

Argue on principle. Roll your eyes at company values all you want — every part of you that thinks values are a little woo-woo is welcome here. But the difference between an org that argues on principle and one that doesn't is stark. They move faster. They generate better results. They argue less. "I'm thinking customer first." "I'm thinking about designing for the customer who finds their way without any help." Oh — we're approaching the same problem from the same point of view. Now we're a team.

Decide first, then experiment. Move the decision from the end of the process to the front. Decide first, gather some evidence, run an experiment on that premise — so you don't have to come back together and debate the same results again.

A few phrases I teach teams to use:

• "How might we…?" A classic, and still gold.
• "What do we already know? What do we need to know?" A great way of stating the obvious.
• "What needs to be true?" My favorite. What needs to be true right now that we can go experiment with, that will tell us we can go direction A, B, or C?

These are the kinds of small language shifts that go in my newsletter, Room to Think.

AI in experimentation — a practical playbook

Katie: Let's get tactical. We're talking about AI, psychological safety, being wrong. AI is exceptionally good at thinking it's right — someone described it as "a teenager: 100% confident and right." How do we balance leveraging AI to speed up the boring stuff while keeping the creativity and the psychological safety?

Rhys: Great question. Experimentation is probably the strongest discipline in marketing making the broadest use of AI right now — beyond just chat windows. It's using just about every flavor of AI tool. Here's how I look at it chronologically, through the innovation cycle.

1. Grab a note-taker. Into your agile meetings, your customer interviews, all of it. I have ADHD — listeners to this pod have probably figured that out by now from the speed I talk — and an accurate AI note-taker is an unlock. Be polite, tell people you're recording them, but take that admin off your plate so you can actually focus on the human in front of you. Then dump the transcript into your knowledge base — also AI-maintained — and use it to find patterns and incongruences across interviews.

2. Use chat (LLMs) to prepare. Create a script on topic. "We can't seem to move customers past this payment page. Here are three transcripts from customers stuck there. What am I missing?" The model will surface things that make you go, "Yeah — hadn't considered that."

3. Run a mock interview with AI. I know we don't like pretending AI is human, but debating Claude — asking it to be a debate partner based on your body of knowledge — is incredible. It puts a mirror to the way you solve problems. "Oh gosh, that's a blind spot. Thanks for raising that." It makes you a better interviewer.

4. Prototyping is where AI is strongest right now. There are different tiers. At the top — what Kameleoon is doing with PBX — clickable, resonant prototypes generated in seconds that actually get people to click and do the thing. That's engineered, that's the high-level stuff. Not every team is there yet, but they should aspire to it.

At the lower end: for some of my outbound, I'll download the original HTML of a prospect's pricing or homepage and ask Claude, "If their customer were more concerned about these things, can you write a better page that reads like that?" Two minutes. That's a prototype too.

Stay in your lane on fidelity. This is the AI pitfall: don't generate a sparkly, brand-new, beautiful clickable prototype right away — you'll have the wrong conversation. When you're talking sketches and wires, you're talking structure and flow. When you're talking static prototypes, you're talking color and placement. When you're on a clickable prototype, you're talking everything. You miss really important conversations if you move too quickly into a high-glossy clickable prototype.

The bigger point: AI isn't there to tell you what it thinks about your current content — that's looking backwards. We're using AI to surface what we haven't seen yet, and to make usable, practical, quick provocations to get a reaction from the market. It's great at making something good enough to get a signal.

And this isn't putting designers out of work. I worry about this a lot. It's creating something faster that designers can then wrap their hands around and do the best work of their lives — because the direction is already validated.

When the room is stuck: follow the premise, not the design

Rhys: Katie, you mentioned: we're in a room, debating the same thing, we've interviewed ten people, why are we still here? I've been there. Here's how we got around it when I was at Intuit, solving really gnarly tax-filer problems — competitive moment, new entrant, deadline pressure.

The way you tie experiments and their success together is by following the underlying premise motivating your buyer. The classics:

• Trust — do they trust the brand, the security of the page? Foundational.
• Usability — can the user actually proceed from screen to screen, find the button, toggle the thing?
• Suitability — does the product solve their problem, and does the messaging say so?
• Then the more "marketery" stuff — FOMO, urgency, "do I see myself in this product?", luxury.

These foundational buyer behaviors are what actually link your experiments together. At Intuit, one of our big wins was helping mobile users compare products more easily. Comparability was the heuristic. So we kept pulling that thread — finding other experiments in other areas of the business that solved for comparability — because we'd already proven the win.

If you're stuck in a room, don't worry so much about design. Think about the effect on the customer. Follow the actual motivating factor behind why they're clicking. Your success rates will shoot through the roof. Your team will be more motivated. You'll win more often. Eventually you'll hit a ceiling, and that's when you go back into the good work and solve a fresh problem.

Psychological safety meets a "confidently wrong" tool

Katie: I love that. But back to AI — it's basically a tool that's confidently wrong. How do you leverage the "rightest in the room" tool while maintaining the "wrongest in the room" culture?

Rhys: Great question. How do you use a tool that is confidently wrong — "60% of the time it works every time" — in a process that's supposed to be rigorous and accurate?

The answer is leaning on the more human skills. That's not a coy answer. For the younger or more junior listener trying to find their place: your cheat code is the soft skills. The things AI is never going to take from you. Here's how that applies to experimentation specifically — and this is real work, not woo-woo.

1. Facilitation. Facilitating a great meeting, interview, brainstorm — so you get results. Reading the room. Never causing offense. Getting quiet people to speak up. You need empathy. You need to change your language, change your rapport. It's a tough skill, and you need practice to learn it.

2. Read the emotions, not just the prompt. AI today isn't good at understanding what's behind the "why". You need a human to understand that an interviewee is hiding something. You need a human to ask politely and kindly if they'd elaborate on a subject that's tough for them. If your product deals with a tricky subject — pain, inadequacy, anything sensitive — you need to read emotions empathetically so the interview can continue. That extends to copywriting too: a prototype provocative enough to get attention, but not enough to put them off the page.

3. Self-correct with kindness. This is on the nose given we're talking about hallucinations. The simplest definition of an experiment is: am I willing to change my position based on new evidence? Yes or no. That's it. Now you're a scientist.

If you can walk into a room with your test card and say, "Okay team — here's the update. Didn't go to plan. That's okay. But here's what we learned, and here's what we're doing next" — that's self-correction. AI is not great at that yet. You have to tell it it's wrong, and then it comes back, "You're so right, I'm so sorry…" We can't operate like that. We can't rely on AI for its own self-correction. That's a human skill. Lean on it.

Speed is the new edge

Rhys: One more piece for the founders who want to move faster: speed is your most important weapon now. This isn't my original idea — it's out there in the Twitterverse — but building used to be the bottleneck. With no-code and vibe-coding, building is no longer the biggest hurdle. Distribution is. Getting your signal and your marketing message into all the noise — that's the hardest part.

The way teams will succeed with AI is by using it to speed up their existing process: run more interviews faster, get stronger signals with more creative ideas in the market faster, facilitate great meetings so ideas come out of brains faster and into market faster. And it's not simple — it takes real leadership and empathy to help a team collaborate, hit a sprint, move that fast. But that's the skill you'll need to win.

What great experimentation leaders (and teams) look like

Katie: Let's talk about leaders who already have a solid AI strategy — they're vibe-coding, they're using PBX, there's a human in the loop, the program is working. What are the traits that tell them they also have the psychological safety needed to create a culture of experimentation with longevity?

Rhys: I'll go further and talk about the traits of their teams too. And I want to be clear with listeners: I didn't come by a lot of these naturally myself. You can learn this stuff. It takes practice, sometimes a little coaching.

Leaders:

Use empathy to set the bar — and hold it. Empathy isn't passive. It means listening, but it also means being firm in your direction and firm about what the result needs to be: the quality, the amount of evidence, the rigor needed to make a decision.

There's still a lot of subjectivity in leadership-level decision-making. That's not going anywhere. But the experimentation team's job is to give their leader the most precise evidence possible to make a difficult, quick decision. So articulate it clearly: "This is what I'm demanding of you. I've got a big decision, it looks like this, I need to make it by this date. I need your help finding evidence that looks like this, this, and this."

Make the trade explicit. "I'll give you autonomy. You can spend some budget. You can take some risks. I trust you to move without my constant input, because you know the goal." On great experimentation teams, that divide gets wider over time — leaders looking ahead at what's next, teams hustling to come back quickly with evidence to feed them. You'll start hand-in-hand and grow out from there.

Teams:

Come into meetings with some level of agreement already. You're not ready to hit the gas if you're still locking horns on fundamentals. You need to be pointing in the same direction. I write about this in my newsletter — there are three layers of goals:

1. The strategic goal (above my pay grade)
2. The ops/team goal ("we're pushing new paid users")
3. The work I see in my own day ("I can do that by getting more engagement on this content")

We all feed into the same thing. If you don't have that yet, leaders, start there. The team needs to see themselves in the work and in the motions carrying it to market.

Look for high participation, not high ownership. If 20% of the people are providing 80% of the ideas, smooth that curve. Get more people contributing. They don't have to own the experiment to participate — they can flag risks, contribute QA, suggest better copy if they're a copywriter. That's their role. If you've got high participation, lots of ideas, and a team challenging the status quo, you're in the right place to take on PBX, take on the craziest new idea, take on Figma Make. You can do that — as long as you're aligned.

The failure: when the three-legged stool tips over

Katie: Being an experimenter means you have a bit of a low ego. The fundamental DNA of what we do is saying "I want to learn, I don't already know." So tell me about a failure you've used to propel yourself forward. Blast it on the internet, publish it forever.

Rhys: Take a number, Katie. Statistically, I think my ideas stink the most out of anyone in any room. I have a lot of ideas — that's part of what I bring — but it comes with a lot of stumbles.

The one I'll dig up was ego-led. And it's why I no longer involve my ego in experiments.

I'll start in the middle. The premise was winning — crushing it across the entire organization. It was a very deliberate reduction: we reduced reading levels on product copy (use your Hemingway app, listeners), reduced the number of features shown, hid things to reduce cognitive load on product pages. (A lot of reductive experiments do very well, by the way — you don't always have to add. Reduction is effective.)

It was crushing in Canada. It was crushing in the US. It was the premise carrying us forward. And Katie — it fed my ego a little bit. I got to walk into executive rooms and show off our winning experiment and double-digit growth.

Here's how not to do the next part. I call it the three-legged stool. If your context changes — the channel, the medium, wherever you're putting it — or if your customer changes fundamentally, or if the content itself, the mechanism you're testing, changes fundamentally, you have to retest. You can't rely on a result being as stable when one of those legs changes.

So I took this winning recipe to another business line. Fundamentally different business — not PLG, more sales-led, very hands-on, different kind of customer. And there I was saying: reduce, cut all your copy, lower the reading level, drop everything.

How do you think it did? It bombed. Embarrassing. I had to walk back into the executive room having promised "this is probably going to win" — and now we were down a couple of points and going back to the original.

What I learned: first, the three-legged stool. Second — you can never oversell an experiment. There are too many ways to embarrass yourself when results don't come out as expected. The best way to behave as an experimenter is to set your range, set your expectations, set the table. State what you believe to be true. It comes out yay or nay. I had a strong sense this one would land — but you live and learn.

Try very hard not to do what I did. Every time you're going to run an experiment, reset your expectations and go, "Okay, this could fail. We need to be ready for the next thing — and a plan for it." I oversold hard.

Katie: I think we've all done that. And then it can undermine your authority — "no, I swear I've been doing this for ten years and it's never happened like this." But it's helpful, you know. Some of the biggest names in experimentation — how do you think they got there? They failed. And as a junior, I remember thinking every idea had to be a winner. It's nice to end with: no, it doesn't. The only way we excel is by failing forward.

What to do tomorrow

Katie: Final question. Someone's listening to this — and if you've made it this far, we love you. What are they doing tomorrow to take action on what you've said today?

Rhys: Watch me go. First, sign up for my newsletter, Room to Think, at ForMentor.ca. That's my consulting practice, helping teams build world-class experimentation organizations. Room to Think is my weekly-ish attempt at sharing three important things every week:

1. One creative power-up — a facilitation activity to run a better meeting and get more ideas out of the room. Lots of little fun activities — there's even one called "What Would Taylor Swift Do?"
2. One small, easy thing you can test in your market tomorrow to get a signal on your customer.
3. One behavioral science nugget — a cognitive heuristic or piece of neuroscience that helps explain why customers behave the way they do. If we can explain that, we're onto something. Maybe you won't make the same mistakes I did. Maybe you'll actually launch a winner.

Second, just get testing. Remove yourself from the experiment. Remove your name from the idea. Go first — put the shaky, not-well-thought-out, rough idea on the wall and see how your team takes to it. Then let them go. Give them some rope, some experimentation runway, let them try and fail, and be gracious when they do. But keep the bar high. Always ask: "Great — we learned something. What are we doing with it next?"

Katie: Rhys, thank you so much for your time. This was incredible. Thank you for being on Unite Voices.

Rhys: Thank you so much for having me. What an honor.