People respond to conflict in patterned ways. Knowing your own habits, and recognizing the stage a conflict has reached, are two of the most useful skills in this field.

Five conflict modes

In the 1970s the management researchers Kenneth Thomas and Ralph Kilmann developed a widely used model, now called the Thomas-Kilmann instrument. It builds on earlier work by Robert Blake and Jane Mouton. The model places responses on two dimensions: assertiveness (how far you pursue your own concerns) and cooperativeness (how far you try to satisfy the other person's). Combining them gives five modes.

• Competing: high assertiveness, low cooperativeness. You push for your own position. It suits emergencies or matters of principle, but used constantly it damages relationships.
• Accommodating: low assertiveness, high cooperativeness. You set aside your own concerns to satisfy the other person. It is useful when the issue matters more to them, or to build goodwill, but can breed resentment if overused.
• Avoiding: low on both. You sidestep or postpone the conflict. It is wise for trivial issues or when tempers need to cool, but it lets important problems fester.
• Collaborating: high on both. You work with the other person to find a solution that fully meets both sets of concerns. It produces the best outcomes on complex issues but takes time and trust.
• Compromising: moderate on both. Each side gives up something to reach a middle ground. It is quick and fair-seeming but may leave real needs unmet.

The key lesson of the model is that no single mode is always best. Each fits some situations. Skilled people notice their default and choose deliberately.

How conflicts escalate

Conflicts tend to follow a recognizable path as they worsen. Issues multiply, the parties stop talking directly, they start seeing each other as enemies, and the goal shifts from solving a problem to hurting the other side.

Glasl's nine stages

The Austrian conflict researcher Friedrich Glasl described escalation in nine stages, grouped into three phases. In paraphrase:

1. Phase one: both can still win. Positions harden, but the parties still believe the matter can be talked through. Debate grows sharper and more tactical. Then they stop believing talk will help and start acting on their own, presenting the other side with facts on the ground.
2. Phase two: one must win and one must lose. The parties form coalitions and fixed images of each other, seeking allies and spreading stereotypes. Then comes loss of face: public attacks aimed at unmasking the other as immoral. Then threats and ultimatums.
3. Phase three: both lose. The parties accept damage to themselves if the other is hurt more. Attacks aim to destroy the other's power base, and at the last stage both sides go “together into the abyss,” willing to be destroyed as long as the enemy falls too.

Glasl's model is not a law of nature; conflicts can skip stages or stall. Its value is practical. It suggests that the right response depends on the stage. In the first phase, the parties can often help themselves through direct conversation. In the second, a neutral third party such as a mediator is usually needed. In the third, outside power, such as courts or peacekeepers, may be required to stop the damage before any talking can happen.

De-escalation begins early

Because each stage makes the next easier, the cheapest time to intervene is early. Small acts, such as raising a concern privately before it becomes a public complaint, or asking a question instead of making an accusation, can keep a conflict in the phase where both sides can still win.

Worked example: the shared office kitchen

In a small office, Priya keeps finding dirty dishes in the shared sink. She suspects Marcus. Track the conflict through Glasl's phases, and notice where a different conflict style could have changed the path.

Phase one. Priya mentions to Marcus, a bit sharply, that people should wash their dishes. He says he always does. Positions harden. A week later she posts a sign: “Your mother doesn't work here.” Talk has given way to action.

Phase two. Priya tells two colleagues it is “obviously Marcus.” They agree, and a coalition forms. Marcus hears about it and feels accused in front of everyone, a loss of face. He replies in a group chat that some people should mind their own business. Priya says she will go to the manager if it happens again: a threat.

Phase three would be reached if, say, each began quietly undermining the other's work, accepting damage to the team to get even.

Now replay it. At the start, Priya was avoiding (saying nothing for weeks) and then switched abruptly to competing. If she had tried collaborating early, she might have said: “I keep finding dishes in the sink and it's getting to me. Can we figure out a system?” It might have turned out that a visiting contractor was the culprit, and a rota or a dishwasher schedule could have solved it.

The manager, arriving in phase two, does what Glasl's model suggests: she acts as a neutral third party, meets each separately, then brings them together to agree on a simple rule.