02 · Conflict Resolution as a Servant Leader¶
Two people on your team stop cc'ing each other. A handover gets done badly twice in a row and both sides describe the other as "impossible." Nobody has shouted, and nothing has been formally escalated — which is exactly why most managers leave it alone and hope it resolves. It rarely does; it goes underground and comes back as attrition. A servant leader's job in team conflict is neither to pick a winner nor to smooth it over, but to restore a working relationship the two people can run themselves. This module gives you a mode-selection grid, a three-meeting mediation structure with actual scripts, and a worked mediation.
1. Five conflict modes, and when each is right¶
The Thomas-Kilmann grid maps behaviour on two axes: how much you push for your own concern, and how much you accommodate the other's.
| Mode | Assertive? | Cooperative? | Right when | Cost of overusing |
|---|---|---|---|---|
| Competing | High | Low | Emergencies, safety, unpopular but necessary calls | People stop telling you things |
| Accommodating | Low | High | You're wrong, or the issue matters far more to them | Your own team's needs get traded away |
| Avoiding | Low | Low | Trivial issues, or emotions too hot to talk today | Problems compound; you get a reputation for absence |
| Compromising | Medium | Medium | Equal-power parties, moderate stakes, time pressure | Both sides half-satisfied, root cause untouched |
| Collaborating | High | High | The relationship is ongoing and the issue is real | Slow; wasteful on trivia |
Two practical points. First, most managers have one default mode and use it everywhere — find yours honestly, because it is the one you use when it is wrong. Second, "avoiding until tomorrow" is a legitimate tactical choice; "avoiding indefinitely and calling it giving them space" is not.
2. Positions vs. interests¶
The single highest-leverage move in any conflict is to get underneath what each person is demanding to what they actually need. Positions conflict almost by definition; interests very often don't.
| Person | Position (what they say they want) | Interest (why) |
|---|---|---|
| Rosa | "Sam has to send me the figures by Wednesday" | She needs to not be blamed on Friday for numbers she got hours before |
| Sam | "I can't commit to Wednesday" | His source data doesn't land until Thursday; he's tired of promising what he can't deliver |
The positions are irreconcilable. The interests — Rosa not being blamed, Sam not over-promising — are entirely compatible, and the moment they're on the table the options appear (partial figures Wednesday with a flagged gap, Rosa's Friday note carrying the timing caveat, escalating the upstream data date). Your job as the mediator is to move the conversation one level down, from what to why.
3. The three-meeting mediation structure¶
Never start with the joint meeting. Putting two angry people in a room with no preparation is how a manageable friction becomes an HR case.
Meeting 1 and 2 — separate, 30 minutes each, same week
Identical structure with each person:
- Frame. "I want to understand your view of what's happening with X. I'm not here to judge who's right, and I'm not going to repeat what you say to me without telling you first."
- Listen for facts. "Walk me through what happened, specifically — dates, what was said." Reflect back (Level 1, Module 3).
- Surface the interest. "What do you need to be able to do your job well here?" Write down the answer in their words.
- Test their empathy. "What do you think this looks like from their side?" A person who cannot answer this at all is not ready for the joint meeting yet — do a second one-to-one first.
- Get consent. "I'd like to get you both in a room to agree how this works going forward. What would you need for that to be useful?"
Meeting 3 — joint, 45 minutes, you facilitate
- You state the purpose and the ground rules (below), not the history.
- Each person gets four uninterrupted minutes to say what they need. The other listens and then reflects it back before responding — that reflect-back requirement does more work than any other single rule.
- You name the shared interest out loud.
- They generate options; you write them down and stay quiet.
- They agree specific, dated commitments. You write these down and send them to both afterwards.
- You set a check-in date, two to three weeks out.
Ground rules to read aloud at the start of the joint meeting
- Speak about behaviour and impact, not character or motive.
- No interrupting; you'll get your turn, and it's timed.
- Before you respond, say back what you heard.
- What's agreed here, I'll write down and send to both of you.
- Nothing said here goes into anyone's performance review.
4. A worked mediation¶
Rosa (financial analyst) and Sam (data operations) have had two months of escalating friction over a weekly figures handover. Their manager, Ken, runs the structure above. Below is the joint meeting.
Ken: Purpose today is one thing: agreeing how the Wednesday handover works from now on. I'm not relitigating the last two months. Ground rules — behaviour not character, no interrupting, and before either of you responds, say back what you heard. Rosa, four minutes.
Rosa: The last three Fridays I've presented numbers I received at 4pm Thursday. Two of those had errors I didn't catch. In the March review the CFO asked me a question about a variance and I couldn't answer it, in front of eleven people. I need enough time to actually check the numbers I put my name on.
Ken: Sam, before you respond — what did you hear?
Sam: That you got the file at 4pm three weeks running, two had errors, and you got caught out in front of the CFO. And that the real issue is checking time, not the deadline itself.
Rosa: Yes. That's it exactly.
Ken: Sam, four minutes.
Sam: I get the source extract from the regional systems Thursday at noon. It has never once arrived Wednesday. When I committed to Wednesday in January I was guessing, and I've been failing that commitment every week since, which is why I stopped replying to the chase emails — not to be rude, I just had nothing new to say. I need to stop promising a date I don't control.
Ken: Rosa, what did you hear?
Rosa: That the Wednesday date was never actually achievable because the upstream extract lands Thursday noon, and the silence on the chases was you having nothing to report, not ignoring me. I genuinely did not know about the Thursday extract.
Ken: So what I'm hearing underneath both of those: Rosa needs checking time before she puts her name on a number, and Sam needs to only commit to things he controls. Those two don't actually fight each other. What could you do?
Sam: I could send what I do have Wednesday — about 70% of the lines are from systems that report Tuesday — and flag the regional block as pending.
Rosa: That would help a lot. I could also add a standing footnote that regional figures are provisional until Friday morning. Then I'm not defending numbers I can't defend.
Sam: And I'd rather we jointly ask whether the regional extract can move to Tuesday. I've never asked. I assumed it was fixed.
Ken: Say the commitments back to me as specific things with dates.
Sam: Every Wednesday 2pm, partial file with pending lines marked. Starting this week.
Rosa: Standing provisional footnote from this Friday. And I chase in the shared channel, not DM, so we both have the history.
Sam: We email the regional data owner together by Friday asking about a Tuesday extract.
Ken: I'll send those three back to you both today. We meet again in three weeks — fifteen minutes, just to check whether it's holding.
The conflict was never about a deadline. It was about Rosa's exposure and Sam's over-commitment, and neither knew the other's constraint. Ken contributed no solutions at all — he supplied structure, the reflect-back rule, and one sentence naming the shared interest.
5. When mediation is the wrong tool¶
- Conduct issues. Bullying, harassment, or discrimination is not a conflict between equals and must not be mediated. Route it to HR, and do not put the affected person in a room to "work it out."
- Persistent bad faith. If one person agrees in the room and does the opposite twice, you have a performance conversation (Module 6), not a mediation.
- Power asymmetry. If one party reports to the other, joint mediation is compromised. Handle it as a management conversation instead.
How It Actually Works¶
The positions-vs-interests distinction (from the Harvard Negotiation Project's principled-negotiation research) works because positions and interests occupy different points in a solution space. A position ("I need the deadline moved") is one specific point already selected from a larger space of solutions that could satisfy the actual underlying interest ("I'm worried about quality if we rush"). Two people arguing their positions are comparing two single points and finding they don't match — which looks like a zero-sum conflict. Once both interests are surfaced, the search moves to the much larger space of solutions that could jointly satisfy both interests (more resources, reduced scope, a phased release), most of which neither party had considered because they'd each already narrowed to one point before the conversation started. Mediation structured around positions is mathematically more likely to find no overlap; mediation structured around interests is searching a strictly larger space.
Why the three-meeting structure (separate, then joint) outperforms jumping straight to a joint conversation. Conflict activates the same threat-response system covered in the empathy and psychological-safety modules — under threat, people's stated positions harden and their ability to generate creative options narrows (a documented effect called "threat-induced cognitive constriction"). A private first meeting lets each party's threat response settle before they have to reason jointly, and it lets the mediator identify each side's real interest without the performative hardening that happens when a position is stated in front of the other party (once someone states a position publicly, backing off it costs face, which makes it harder to move even after they'd privately concede the point).
Why choosing the wrong conflict mode is a mismatch cost, not just a preference issue. Each of the five conflict-handling modes (Thomas-Kilmann: competing, collaborating, compromising, avoiding, accommodating) trades assertiveness against cooperativeness differently, and each is the objectively correct choice for a different structural situation — a one-time, low-stakes disagreement genuinely doesn't warrant the time cost of full collaboration, while a high-stakes recurring disagreement between people who'll keep working together genuinely does. A leader defaulting to one mode regardless of situation (commonly: always avoiding, or always competing) is applying a fixed strategy to a variable-payoff game, which guarantees suboptimal outcomes in every situation that doesn't match their default.
Exercise¶
Identify one live friction on your team — including the mild kind you've been telling yourself will sort itself out. Write down each person's position in one sentence, then their probable interest underneath it. If you can't state either interest with confidence, that alone is your evidence that you need the two separate meetings.
Then run at least Meeting 1 and 2 from section 3 this week, using the five-step structure verbatim. Afterwards, note: did the interests turn out to be genuinely incompatible, or just the positions? And which of the five conflict modes in section 1 were you personally reaching for before you started?