Everyone says leaders need self-awareness. Nobody hands you the manual.
If you came up through engineering, the human side of the job can feel like undocumented code running in the background. It clearly does something. You just can't read the source. This is the approach that finally made it readable for me: treat self-awareness as a system you can debug.
Finding the API docs
For years I ran a set of mental processes without ever naming them. They worked quietly in the background of every interaction. When I finally had labels for them, it felt like finding the API docs for a tool I'd been using for a decade.
Once something has a name, you can see it. Once you can see it, you can get curious about it, and then improve it. Self-awareness stopped being a vague virtue and became a skill with inputs, outputs and a feedback loop.
I think of it as three sets of variables:
- External: how I interact with other people
- Internal: how I interact with myself
- Context: how I interact with the environment I'm in
Self-awareness is the ongoing work of updating those variables as new feedback arrives. It isn't a trait you either have or don't.
The research backs this up. Organizational psychologist Tasha Eurich's studies suggest that about 95% of people believe they're self-aware, but only 10–15% actually are. Her two dimensions, internal (knowing yourself) and external (knowing how others see you), map neatly onto the variables above.
Why leadership exposes the bugs
You notice all this most when you start leading, because now you're operating the human interfaces from the manager's seat. The first attempts rarely behave as expected. That's when you find out the model of the world in your head isn't the one running in everyone else's.
I learned it by walking into rooms with the wrong assumptions. Once I met with a group of non-technical colleagues, assuming they shared my curiosity about how information flows through systems. I tried to pull them into it, got pushback, and reacted in a sequence I now recognize:
- Surprise: "Don't you know this?"
- Mild judgment: "Why didn't you take the time to learn it?"
- Correction: "Oh. You have other priorities, and you're good at things I'm not."
- Genuine curiosity: "What are your priorities, and how do you actually make decisions?"
Only the fourth reaction was useful. Reading what was actually in other people's heads turned out to matter more than any amount of introspection about my own.
Your model of a person is a map with blank regions
A note for the logic-first reader, which is probably you. What follows models people almost like systems, while my whole point is that people are not systems. Both are true, and the tension is where the value is.
A mental model of someone is never an accurate picture of how they think. It's scaffolding. Its job is to show you where your data is missing. Even a perfect map of someone's brain would miss their current state: the stress, the bad night's sleep, the thing their manager said an hour ago.
So use the model for one thing: flagging the blank regions.
The four-question routine
Before an important conversation, I run four questions. They matter most when there's a conflict to defuse, but they pay off in calm moments too.
- My model of me. How do I see this situation, and what outcome do I want?
- My model of them. What do I think they believe, and what do they want?
- Their actual model. I ask them those two questions directly instead of guessing.
- Their model of me. How do they think I see this, and what do they assume I want?
The fourth question is external self-awareness in practice: checking how you actually land, not how you hope you land.
The useful signal is the diff between questions 2 and 3: my guess about them versus their real answer. When they differ, I don't treat either side as wrong. I acknowledge the gap, ask clarifying questions and build a bridge. The goal isn't to change their mind. It's to find an analogy or a process that lets both sides keep working.
A real example. Someone on a team was doing poorly on a technical task. My model said it was a skill gap or an effort problem. Their actual model, once we talked, was that they had never wanted to work in tech. They wanted a client-facing role. What I'd read as a performance issue was a person in the wrong seat, and only the conversation surfaced it.
Reading the room, and reading yourself
Some of my hardest lessons weren't about other people's models at all. They were about misreading signals right in front of me.
Good data, bad inference. I have 20/15 eyesight. As a student I'd spot someone across campus, wave and get nothing back, so I decided those people didn't like me. It took me years to realize they simply couldn't see me from that far away. My eyes produced accurate data, and my inference got it completely wrong.
People prefer different interfaces. An engineer I worked with rushed past me in hallways and barely acknowledged me at the gym. Online, we had great exchanges: constant chats, emojis, images back and forth. In-person was just an uncomfortable channel for them. So I stopped pushing that channel and met them where they communicated best.
Know when the room isn't yours. Early in my career I supported a trading desk. At my first company, tech and business people were on a first-name basis and engineers were welcome in business meetings. At the next bank I carried that assumption straight in, joined a huddle about a trading strategy and offered my opinion. Silence. I hadn't even realized I was expected to leave until someone told me. One company's culture doesn't transfer to another.
Control your body, not your face. That episode made me much more aware of how I come across, but I didn't want to suppress my reactions into something fake. I can't fully control my face, and I don't want to. What I can control is my posture: keeping it open on purpose, so information keeps flowing instead of the loop closing early on a guess.
And go easy on first impressions. People's expressions aren't always under their control, and they aren't always about you. It takes several interactions to understand someone. Notice the feeling a first meeting gives you, but hold off on the verdict. If it helps, ask a peer something exploratory: "How's your working relationship with them?" or "What do they do well?"
For the skeptical engineer
Plenty of tech leaders pride themselves on logic over soft skills. I share the instinct, so here's the honest version.
None of my communication classes were mandatory in my engineering education. My whole idea of communication came from engineering books, not from how communication actually works between people. We're trained to solve problems with the human context stripped out, and that quietly teaches us the human context doesn't matter.
In practice it's at least half the outcome, often more. If I'd understood that earlier, I would have taken the human interfaces as seriously as the technical ones, and seen how much they shape the engineering itself.
The map of the person across the table is never complete. Knowing where it's blank, and being willing to ask, is most of the job.
TL;DR
- Self-awareness is a learnable skill. Treat it like a system you can debug.
- Track three sets of variables: external (others), internal (you) and context (environment).
- Your model of another person is a map that flags blank regions, not a picture of their mind.
- Before high-stakes conversations, run the four questions, then close the gap between your guess and their real answer by asking.
- One interaction is one data point. People prefer different interfaces.
I'm Vladimir Baranov, an executive coach for tech and science executives and founders, and previously co-founder and CTO of AdvisorEngine. This post first appeared on Human Interfaces.
Top comments (0)