Most managers have one conflict move they make without thinking. Some smooth things over before anyone has finished talking. Some go quiet and hope it passes. Some plant a flag and defend it. The trouble isn't the move itself, it's that it's the only move, applied to a leaking tap and a burst pipe alike. The Thomas-Kilmann model exists to turn that reflex into a choice.

The quick version

  • Kenneth Thomas and Ralph Kilmann (1974) mapped conflict behaviour on two axes, assertiveness (pushing for your own concerns) and cooperativeness (caring for the other side's).
  • That grid produces five modes: competing, collaborating, compromising, avoiding, accommodating. None is "best"; each fits some situations and misfits others.
  • The skill leaders are missing is rarely a new mode, it's the judgement to match the mode to the stakes and the relationship, and the range to switch when needed.
  • Treat the model as a shared vocabulary for naming what's happening, not as a scorecard of how good you are.

The idea in depth

Thomas and Kilmann's claim is deceptively simple: any conflict behaviour can be located using just two questions. How hard are you pushing for what you need (assertiveness)? And how much weight are you giving what they need (cooperativeness)? Cross those two and you get five recognisable modes. Competing is high-assertive, low-cooperative, winning. Accommodating is the mirror, low-assertive, high-cooperative, letting the other side have it. Avoiding sits low on both, sidestepping. Collaborating is high on both, the genuine win-win that satisfies everyone's concerns. And compromising sits in the middle of both axes, a quick split-the-difference deal that partly satisfies each side. The instrument itself isn't a personality quiz with right answers; it is thirty pairs of statements deliberately balanced so neither option sounds more flattering, which is meant to surface what you actually do rather than what you'd like to claim (Thomas & Kilmann, 1974; Kilmann Diagnostics; overview).

quadrantChart
  title The five conflict modes
  x-axis "Low cooperativeness" --> "High cooperativeness"
  y-axis "Low assertiveness" --> "High assertiveness"
  quadrant-1 "Collaborating"
  quadrant-2 "Competing"
  quadrant-3 "Avoiding"
  quadrant-4 "Accommodating"
  "Compromising": [0.5, 0.5]
The two axes, assertiveness and cooperativeness, and where each mode sits. Compromising lands in the middle of both. Leaders Loop

The two axes didn't appear from nowhere. The model is a refinement of Robert Blake and Jane Mouton's 1960s managerial grid, which plotted "concern for people" against "concern for task"; Thomas reframed those into assertiveness and cooperativeness for the specific case of conflict (Kilmann Diagnostics, history). So the move is: after your next tense exchange, replay it on those two axes. Were you pushing for your concern, theirs, both, or neither? You will usually find you defaulted to a habit rather than reading the room, which is exactly the gap that reading the room is meant to close.

No mode is "best", each has a home situation

The most common misreading of this model is to crown collaborating the winner and file the rest under bad habits. Thomas and Kilmann are explicit that all five are legitimate, and that even the unfashionable ones have a job. Competing is right when a fast, unpopular call has to be made, a safety issue, a deadline that can't move. Accommodating is right when you're wrong, or when the relationship matters more than this particular point. Avoiding is right when the issue is trivial or the heat is too high for anyone to think straight. Compromising is the pragmatic middle when time is short and the stakes are moderate. Collaborating is for the issues important enough to justify the work it takes (The Myers-Briggs Company, current publisher of the instrument).

The skill isn't having a favourite mode. It's not having one when the situation calls for a different one.

So the move is: name your default out loud, to yourself, or to a trusted peer. "Under pressure I accommodate" or "I compete when I feel rushed." Once the reflex has a name, you can catch it mid-conflict and ask the only question that matters: does this situation actually want that? Widening your range is what relationship management looks like in practice, handling the same person differently depending on what the moment needs.

What the evidence does, and doesn't, support

Here is the honest limitation, and it cuts two ways. First, the instrument measures self-reported preference, not demonstrated skill. Knowing you favour collaborating tells you nothing about whether you're any good at it. Worse, because the modes are easy to learn, people game the questionnaire once they've seen it a few times, a criticism noted as far back as the early 1990s, when one columnist admitted he'd taken it so often he knew which answers produced the result he wanted (Thomas-Kilmann Conflict Mode Instrument, Wikipedia). Treat your score as a conversation starter, not a verdict.

Second, the model says nothing about a deeper finding: not all conflict is equal. De Dreu and Weingart's meta-analysis of 35 studies (Journal of Applied Psychology, 2003, 88(4), pp. 741–749) found that both relationship conflict and, surprisingly, task conflict were negatively associated with team performance and member satisfaction, with the damage worse on complex work (De Dreu & Weingart, 2003). The popular advice that "task conflict is healthy" did not hold up cleanly. So the move is: before you pick a mode, separate the what from the who. A disagreement about the plan (task) and a simmering grudge (relationship) need different handling, and the second will poison the first if you let them blur. The skill of telling them apart leans on empathy and social awareness.

A worked example

Consider Priya, an engineering manager, and two reports, Sam and Dani, locked over which of two database designs to ship. (Names and figures here are illustrative.) The deadline is ten days out. Sam's design is faster to build; Dani's scales better but adds, on her own rough estimate, around two weeks. Priya's reflex is competing, she has an opinion and she's senior, but she stops and runs the situation through the grid.

flowchart TD
  A(["A disagreement lands
on your desk"]) --> B{"How high are the
stakes for the outcome?"}
  B -->|"Low / trivial"| C(["Avoid or accommodate,
don't spend the capital"])
  B -->|"High"| D{"How much does the
relationship matter?"}
  D -->|"A lot, and there's time"| E(["Collaborate,
dig for the real interests"])
  D -->|"A lot, but time is short"| F(["Compromise,
a fair, fast split"])
  D -->|"Less than a fast, firm call"| G(["Compete,
decide and own it"])
A rough triage: stakes first, then relationship and time. Illustrative, not a rule. Leaders Loop

The stakes are high, this design will live for years, so avoiding is out. The relationship matters and there's just enough time, so she chooses collaborating over competing. In a 40-minute session she does the one thing collaborating actually requires: she gets underneath the positions. It turns out Sam isn't wedded to his design, he's worried about the deadline. Dani isn't wedded to scale for its own sake, she's been burned by a rewrite before. Those are interests, not positions, and they're not in conflict. The team ships Sam's faster design now, with a documented migration path to Dani's approach booked for the next quarter. Both concerns are met. Had Priya been genuinely out of time, the honest move would have been compromising, ship a hybrid, accept it pleases no one fully, or competing and owning the call. The grid didn't make the decision. It stopped her making it on autopilot.

Frequently asked questions

Is there a single best conflict style?

No. The five modes are a repertoire, not a ranking. Collaborating produces the fullest outcomes but costs the most time and goodwill to pull off; the other four each have situations they fit better. The skill is matching the mode to the stakes and the relationship, not defaulting to a favourite.

What is my conflict style, and can I change it?

Most people lean on one or two modes under pressure. That's a habit, not a fixed trait, and you can widen your range with practice. The first step is noticing, after the fact, which mode you reached for, then asking whether the situation actually called for it. This is ordinary reflective practice.

Isn't avoiding always the wrong answer?

No. Avoiding is the right call when the issue is genuinely trivial, when emotions are too hot for a useful conversation, or when you have no standing to change the outcome. It only becomes a problem when it's a reflex rather than a choice, so real issues never get raised at all.

What's the difference between compromising and collaborating?

Compromising splits the difference: both sides give something up for a quick, acceptable deal. Collaborating digs beneath the stated positions to the interests underneath, aiming for a solution that satisfies both fully. Compromise is faster; collaboration is more complete but slower and harder.

Does the Thomas-Kilmann model actually hold up?

The two-axis map is a useful, widely-taught lens, but it describes self-reported preference rather than proven effectiveness, and people can game the questionnaire once they know how it scores. Use it as a vocabulary for talking about conflict, not as a measurement of how skilled you are at it.

Related in the Toolkit

Where to go next