Resolving Conflict Between People

Managing an Underperformer was about one person and a gap in their work. This page is about two people and the friction between them — the designer and the engineer who can’t agree, the two nurses who’ve stopped speaking on shift, the shift lead and the new hire who each think the other is impossible. When that friction lands on your desk, your instinct pulls in two wrong directions at once: stay out of it and hope it blows over, or hear one side, decide who’s right, and lay down the law. Both feel like management. Both usually make it worse.

This page gives you a third option — the one that actually works. You step in, but not as a judge handing down a verdict. You step in as the person who owns the process by which two people work their own conflict out. That’s a specific skill, and it starts with a distinction most managers never make.

Not all conflict is a problem

The first mistake is treating conflict as a single bad thing to be stamped out. It isn’t. There are two kinds, and they need opposite responses.

Task conflict is disagreement about the work — which approach is better, whether the deadline is realistic, how to prioritize, what the customer actually needs. This kind of conflict is not just tolerable; it’s a sign of a healthy team. Two people arguing hard about the right design, the right treatment plan, the right layout for the warehouse floor are doing exactly what you want them to do. A team with no task conflict isn’t harmonious — it’s usually one where people have stopped caring, or are too afraid to disagree.

Relational conflict is disagreement about each other — “he’s condescending,” “she never listens,” “I can’t stand working with him.” This kind is corrosive. It doesn’t sharpen the work; it poisons the ability to do work together at all. Left alone it spreads: to the people around them, to meetings, to the mood of the whole team.

The manager’s job is not to eliminate conflict. It’s to protect the task conflict and resolve the relational conflict — and, crucially, to notice when one is masquerading as the other. Very often a “disagreement about the work” that won’t die is actually a relational conflict in disguise: they’re not really fighting about the API or the roster, they’re fighting because one feels disrespected by the other.

TASK conflict          →  keep it. Channel it. It's the work working.
   "I think we should ship Friday, not Monday."

RELATIONAL conflict    →  resolve it. It's eating the team.
   "I can't work with someone who talks over me in every meeting."

The trap: relational conflict wearing a task costume.
   Endless, heated re-litigation of a small decision usually
   isn't about the decision.

The common trap — trying to make everyone comfortable

Many managers define success as “no more tension” and rush to smooth every disagreement over. But a smoothed-over team where nobody argues about the work is not a healthy team — it’s an underperforming one that has learned conflict is unwelcome. Your goal is not comfort. It’s a team that can disagree about ideas fiercely and work together warmly. Kill the personal friction; keep the professional heat.

Why avoiding and picking a side both fail

When relational conflict lands on you, the two easy exits are avoidance and adjudication. Understand precisely why each fails, because the failure explains the alternative.

Avoidance — hoping it resolves itself — fails because unaddressed relational conflict almost never dissipates; it hardens. Each person accumulates evidence for their story about the other. They recruit allies. They route around each other, which slows the work and forces everyone nearby to manage the awkwardness. And your silence sends a message: this is acceptable here, and no one is coming to help. The team learns you won’t step into hard things — which quietly lowers their trust in you for everything else.

Picking a side — hearing one version, deciding who’s right, and ruling — fails for a subtler reason. Even when you guess correctly about the facts, you’ve now created a winner and a loser. The “loser” doesn’t change their mind; they conclude the manager plays favorites and the process is rigged. The “winner” learns that the way to resolve conflict here is to get to the boss first with the better story. You haven’t ended the conflict; you’ve taught people to litigate it upward. And you’ve taken ownership of an outcome that was never yours to own.

The alternative is to separate two things the easy exits confuse: the process and the verdict. You own the process — you convene the conversation, set the rules, keep it safe and honest, and make sure a real agreement gets reached. You do not own the verdict — the content of the agreement belongs to the two of them wherever possible. A mediator, not a judge. This is the single most important idea on the page.

The evidence — psychological safety and disagreement

Google’s multi-year study of its own teams (Project Aristotle, reported publicly around 2015–2016) found the strongest differentiator of effective teams wasn’t talent or seniority but psychological safety — the shared sense that you can speak up, disagree, and admit mistakes without being punished. Read that carefully: safe teams have more open disagreement, not less. The takeaway for conflict isn’t to suppress friction; it’s to build the safety that lets friction stay about ideas instead of curdling into fear and resentment.

Get past positions to interests

Here is the core move that makes mediation possible. In any conflict, each person shows up with a position — the thing they say they want. Underneath the position is an interest — the need, fear, or goal the position is trying to serve. Positions clash. Interests, more often than not, overlap. Your job is to dig from the loud, incompatible positions down to the quieter interests, because that’s where agreement lives.

POSITION (stated)          INTEREST (underneath)
"You have to ship it   →   "I promised the client Friday and I'm
 by Friday."               scared of looking unreliable to them."

"It's not ready, I     →   "If we ship a bug with my name on it,
 won't rush it."           I'm the one who gets blamed. I need cover."

Look at what happens when the positions become interests. “Ship Friday” versus “not ready” is a wall — there’s no compromise between yes and no. But “I need to look reliable to the client” and “I need cover if it breaks” are not in conflict at all. There’s a dozen agreements that serve both: ship Friday with a clearly communicated known-issues list, ship a limited version Friday and the rest Monday, get the client to agree Monday is fine. The solution was invisible while they argued positions and obvious once they named interests.

The manager’s questions that do this digging are simple and you should memorize them: “What are you actually worried will happen?” “What would you need to feel okay about this?” “What’s the outcome you can’t live with?” These questions move people from defending a stance to describing a need — and a described need is something the other person can often meet without feeling like they lost.

In practice — the roster that wasn’t about the roster

On a hospital ward two nurses were locked in a running fight over the shift roster — who got which weekends. The manager kept trying to solve it as a scheduling puzzle and it kept re-erupting. When she finally asked each of them privately “what are you actually worried about?”, the interests surfaced: one was a single parent terrified of being stuck without childcare on a surprise weekend; the other felt the first one always got accommodated and she was invisible. The roster was the costume. The real agreement — predictable notice for one, explicit acknowledgment and a fair rotation for the other — took ten minutes once the interests were on the table. Two weeks of “scheduling” had never touched it.

A structure for the mediated conversation

Owning the process means running it deliberately, not just getting everyone in a room and hoping. Here’s a structure that works.

1. Separate sessions first, if the heat is high

If the two can’t yet be in a room without it exploding, start by meeting each one alone. The goal of these is not to gather ammunition or take a side — it’s to let each person feel heard, to surface their interests (using the questions above), and to lower the temperature enough that a joint conversation can be productive. You’re also checking for one thing: is this a genuine two-way conflict, or is one person being mistreated? Those are different situations, and you need to know which you’re in before you convene a “let’s work it out together” meeting.

2. Set ground rules out loud

When you bring them together, name the rules before anyone speaks. Something like:

- We're here to solve a problem, not to win.
- One person speaks at a time. No interrupting.
- Talk about behavior and impact, not character.
  ("When the meeting ran over, I couldn't..." not "You're selfish.")
- Everything said here stays professional and stays here.
- We don't leave until we have a concrete next step.

Stating the rules out loud does real work: it makes the process feel fair and safe, and it gives you something neutral to enforce (“hold on — one at a time”) without taking sides.

3. Each side is heard, fully, once

Let each person say their piece without interruption. Then — and this is the move that changes the room — have each one reflect back what they heard the other say, to the other’s satisfaction, before responding. People soften dramatically when they feel genuinely heard, and most relational conflict is fueled by both people feeling unheard. You’re not asking them to agree. You’re asking them to demonstrate they understood.

4. Move to interests, then to a concrete agreement

With both positions on the table, steer toward interests: “So it sounds like what you each really need is ___ and ___. Where do those overlap?” Then push for something specific and behavioral. Not “we’ll communicate better” — that’s a wish, not an agreement. Instead: “I’ll give you the numbers by Wednesday; you’ll flag concerns in the meeting, not after.” Write it down. Set a date to check back. A vague peace is a conflict on pause.

Case study — Toyota’s andon cord

On Toyota’s production lines, famously refined through the decades after the 1950s, any worker can pull the andon cord to stop the line when something’s wrong. What’s instructive for conflict is the response it’s designed to trigger: not blame for the person who stopped the line, but a team gathering to solve the problem at its source. The lesson generalizes. When you build a system where surfacing a problem is treated as a contribution rather than an accusation, problems get raised early and solved cheaply. When surfacing a problem gets you punished, people stay silent and small frictions grow into the corrosive kind. How your team handles conflict is downstream of whether raising it is safe.

When mediation isn’t enough

Mediation is your default, not your only tool. Sometimes you have to stop mediating and act. Know the three cases.

When there’s a genuine deadlock, decide. If the conflict is really about a work decision and, after honest discussion, the two still can’t agree — you decide. This isn’t picking a side in a personal fight; it’s doing your job. Hear both, explain the interests you’re weighing, make the call, and own it clearly: “I’ve heard you both. We’re going with X, for these reasons. I need you both behind it now.” Teams can accept a decision that went against them far more easily than they can accept endless limbo.

When a boundary is being crossed, set it. If one person’s behavior — talking over people, taking credit, sniping — is the actual problem, mediation between “equals” is the wrong frame, because they aren’t in a symmetric disagreement. That’s a performance and conduct conversation with one person, run the way managing an underperformer describes: name the behavior, its impact, and the required change. Don’t launder a conduct problem through a “let’s both compromise” mediation.

When it’s a values violation, escalate. Harassment, discrimination, bullying, safety violations, dishonesty — these are not conflicts to be mediated and they are not compromises to be brokered. Asking a harassed person to “meet the other halfway” is itself a harm. These go to the boundary you don’t negotiate: they stop now, and depending on severity you involve HR, senior leadership, or formal process. Mediation is for two people who both want to work together and can’t. It is never the tool for protecting someone from mistreatment.

How it connects

Resolving conflict draws on nearly everything before it. It rests on the psychological safety and trust you build long before any fight (Part 9 · Hard Situations sits on that foundation). It uses the same behavior-and-impact language as managing an underperformer when a conflict turns out to be a conduct problem. And staying calm in the room — not getting pulled into the emotional weather of the two people in front of you — is the same steadiness you’ll need in handling a crisis calmly.

Try this

Think of one live friction on your team right now — even a low-grade one where two people just seem to grate. This week, meet each of them separately for fifteen minutes and ask only the interest questions: “What are you actually worried will happen? What would you need to feel okay about this?” Don’t propose solutions, don’t mediate yet, don’t take a side. Just write down each person’s underlying interest in one sentence. Then look at the two sentences and ask: where do these overlap? You’ll usually find the overlap is bigger than the fight suggested.

Reflect

1. Think of a recent conflict on your team. Was it task conflict, relational conflict, or a relational conflict wearing a task costume? How can you tell?
2. When conflict lands on you, which easy exit do you reach for first — avoiding it or picking a side? What does that cost you?
3. Recall a time two people were stuck on clashing positions. What were the interests underneath, and did anyone ever surface them?
4. Do the people on your team experience raising a problem as a contribution or as an accusation? What in your reactions taught them that?
5. Where’s the line, for you, between a conflict to mediate and a boundary or values violation to act on directly? Have you ever mediated something you should have simply stopped?

Show reflections

1. The tell is usually whether the disagreement is about the work or about the person, and whether it resolves once a decision is made. A “work” argument that keeps re-erupting after every resolution is almost always relational underneath — that’s the costume to look for.
2. Most managers have a default flinch. Avoiders let friction harden and teach the team no help is coming; side-pickers create winners and losers and teach people to litigate upward. Naming your default is the first step to catching it.
3. The goal is to feel, in a real example, how positions clash while interests often overlap. If nobody surfaced the interests, that’s the missed move — and the reason the conflict felt unsolvable when it usually wasn’t.
4. This is about the psychological safety underneath all conflict. If your first reaction to a raised problem is irritation or blame, you’re training silence — and silence is where small frictions grow into corrosive ones. The andon-cord lesson is that raising a problem should be treated as a gift.
5. A strong answer draws the line clearly: mediation is for two people who both want to work together and can’t; it is never the tool for a conduct problem or a values violation, where asking the harmed person to “meet halfway” is itself a harm. If you’ve mediated something that should have been stopped, that’s the most important lesson on the page to carry forward.