Introduction

On Sunday mornings, I have a ritual. Coffee strong enough to negotiate with Monday, a notebook that swears it’s going to behave, and a tiny promise to myself that I’ll chip away at my future book, It’s a Great Day for Discovering Why. Week in and week out. Persistently. Like a woodpecker with a mission, or me trying to convince myself I only need one more browser tab.

Lately, my rabbit hole has been understanding the science of forming opinions more holistically. Not the loud kind you see on cable news, but the quieter kind that shows up in conference rooms, product meetings, strategy sessions, and yes, the answers a participant provides to our surveys and interviews. The kind that decides whether your best insight becomes action or becomes “interesting” and then gets filed into the corporate basement next to the 2019 brand tracker.

Here’s what I keep bumping into: opinions are not statues. They’re more like stew. They simmer. They pick up flavor from whatever is already in the pot, and they thicken or thin depending on what new ingredients get tossed in. Prior beliefs, fresh evidence, personal goals, and the social weather of the room all blend together.

And that is the real reason “perfect data” can still lose to one stubborn anecdote.

Your brain loves shortcuts, because it’s busy trying to keep you alive

We like to think we reason like engineers. Input, analysis, output. But most days, our brains reason like busy parents at a supermarket checkout with two kids and a dying phone battery. We take shortcuts.

A shortcut is just the brain’s way of saying, “I don’t have time to solve the full math problem, but I still have to decide.” Cognitive scientists call these shortcuts heuristics. I call them “mental autopilot.”

Three of the greatest hits show up over and over:

Availability. If you can easily remember an example, you think it’s more common than it is. One vivid customer complaint can feel like a full market shift.

Representativeness. If something looks like a pattern you recognize, you assume it is that pattern. “This reminds me of that failed launch, so it must be risky.”

Anchoring. The first number you hear can become sticky, even if it’s nonsense. Someone says, “This will cost $2 million,” and suddenly every budget conversation is orbiting that number like it has its own gravitational pull.

If you’ve ever watched a room argue over a forecast like it’s a sports replay, you’ve seen anchoring in the wild. Everyone is smart. Everyone is experienced. And yet the first number still behaves like it’s glued to the whiteboard.

The mind’s favorite hobby is being right

Now we arrive at my least favorite truth, mainly because I can see it in myself.

Once we start leaning toward a belief, we do not objectively shop for evidence. We shop for evidence like we’re trying to justify a purchase we already made. We look for reasons we were right to begin with.

That’s confirmation bias, which in plain English means: we notice, interpret, and remember information that supports our existing story, often without realizing it.

This is how a perfectly rational person can say:

• “That data is interesting, but I’m not sure it applies here,” and then immediately follow with:
• “My friend at Company X tried something similar and it was a disaster.”

One point is evidence. The other point is a campfire tale. Guess which one tends to win when the room is stressed and the decision is risky.

Motivation is the steering wheel

Here’s where it gets spicy, and where “just bring more data” often fails.

People don’t only reason to find the truth. People reason to reach a conclusion that feels acceptable to them. Sometimes they’re motivated by accuracy. Sometimes they’re motivated by protecting a prior decision, an identity, a political position, or their own risk exposure.

This is motivated reasoning, and it explains why “resistance to evidence” is often not about intelligence. It’s about incentives. It’s about fear. It’s about who gets praised, who gets blamed, and who gets stuck holding the bag if the decision goes sideways.

Translation for insight people: the stakeholder may not be rejecting your finding. They may be rejecting the consequences of believing your findings.

That’s a very different problem to solve.

Not everyone processes your message the same way

Ever deliver the same insight to two leaders and get two completely different reactions?

One person says, “This changes everything.” The other says, “Can you put this on one slide, and also can it be less… complicated?”

That’s not just personality. That’s attention and motivation.

When people care deeply and have time, they scrutinize arguments. They want to see assumptions, trade-offs, and proof. When they’re overloaded, distracted, or the topic feels distant, they lean on cues. Is this clear? Is the messenger credible? Does this fit what I already believe?

Same evidence. Different processing mode.

It’s like cooking for two guests. One wants a five-course tasting menu and asks about the origin story of the salt. The other wants a grilled cheese and is offended that you even mentioned truffle oil.

Opinions are social. Teams have gravity.

Now zoom out from the individual brain to the group.

In organizations, opinions don’t just form in heads. They form in networks. People watch what the respected leaders think. They watch what gets rewarded. They notice what gets mocked. And they adjust, often subconsciously.

If the social system is healthy and trust is shared, groups can converge toward a common view. If trust is broken or the camps are far apart, groups can polarize or fracture into separate “local realities,” each with its own narrative that feels reasonable inside that bubble.

This is why you can put the same chart in front of ten people and end up with three different interpretations, two silent resisters, and one guy who asks about the sample size like it’s his emotional support animal.

Sometimes the barrier is not evidence. It’s the trust network.

The jump from opinion to action is not automatic

Here’s the part that matters most for anyone delivering insight: even if you successfully change someone’s opinion, that does not guarantee action.

Action usually shows up when three things line up:

• Attitude: Do I think this is a good idea?
• Norms: Do the people around me support this?
• Control: Can we actually do it, given budgets, capabilities, timelines, and politics?

This is the hidden reason so many insight decks get applause and then die quietly.

A deck can pump up attitudes and still ignore norms and control. You can get agreement without motion.

If you want movement, you have to reduce friction. You have to make the action feel doable. You have to answer the questions nobody asks out loud:

• Who else is going to back me if I say yes?
• What will this cost me personally if it fails?
• What will we stop doing to make room for this?

People don’t just decide based on what’s true. They decide based on what feels safe to own.

Insight is not a chart. It’s a belief update packet.

This is the line I’ve been scribbling in the margins while working on the book:

The most valuable thing we deliver isn’t a finding. It’s a belief update that people can use.

A belief update packet does three things:

• It names the current assumption the organization is operating under.
• It shows the evidence that should update that assumption.
• It makes the implication usable in the real world, with all its constraints.

If you skip that last part, you’re basically handing someone a beautifully wrapped gift with no way to open it.

One of the strongest habits here is to build in disconfirmation. That’s a fancy word for a simple move: show what could have proven you wrong. Include at least one alternative explanation you tested. Explain why you didn’t just build a case, you tested reality.

That earns trust. It also disarms the “You’re just trying to sell us something” reflex.

A quick note on synthetic data, flight simulators, and confidence

Since I’m already deep in the “future of insight” part of the book, I can’t ignore the rise of synthetic data.

Synthetic data is generated data that mimics real data. It can speed up iteration and, with governance, potentially reduce privacy risk. It can be a great sandbox for testing ideas and pipelines.

But it cannot create new ground truth. It can only reflect what the generator learned. That means it can preserve bias, smooth away rare behaviors, or accidentally inflate confidence if you treat it like the real thing.

So here’s the analogy that sticks for me: synthetic data is a flight simulator.

• Fantastic for practice, training, stress-testing, and running scenarios.
• Not the same as flying in real weather with real turbulence.

Use it wisely. Validate against reality when the decision is high-stakes.