The strategy to employ is collaboration, also called the problem-solving or integrative approach: both teams put their positions on the table as interests, jointly generate options, and agree on criteria before they choose. It only works under one condition—each side must hold something the other needs and neither can get its outcome by acting alone. If that condition fails, collaboration turns into a slow stalemate.
The Thomas–Kilmann framework, developed by Kenneth Thomas and Ralph Kilmann in the mid-1970s and still the default conflict-mode instrument in management training, sorts behavior into five modes along two axes: how hard you push for your own concern, and how much you attend to the other side's. Collaboration scores high on both. That matters for your situation, because a win/win outcome is literally the definition of that quadrant.
Imposed win/win is a contradiction in terms. If a project manager announces the "win/win solution" in a kickoff and asks both teams to implement it, what's actually being used is forcing dressed in collaborative language. Collaboration requires that the parties build the solution themselves, or at least recognize it as theirs.
Keep one caution from the empirical work on conflict: Kathleen Eisenhardt and Laurie Weingart found that disagreement about how to do the work improves team performance, while disagreement about who gets what—relationship, status, territory—degrades it. Before you choose a mode, decide which kind of conflict is in front of you. A resource fight and a design dispute need different handling even though both live inside the same project.
What the TKI Five Modes Actually Measure
The instrument's two dimensions determine everything else. Assertiveness is the degree to which you try to satisfy your own concerns. Cooperativeness is the degree to which you try to satisfy the other party's. Collaboration sits at the top of both. Compromise sits in the middle of both. Avoidance is low on both.
The scores are situational, not personality scores. Thomas and Kilmann are explicit that everyone can use all five modes and that the instrument measures tendencies, not fixed types. A team that "always collaborates" is not virtuous; it burns meeting hours on decisions that deserve a quick call.
Each mode carries a real cost. Collaboration consumes the most time and the most emotional labor of the five because it requires genuine disclosure of interests—the reasons behind the demand, not the demand itself. If either team's representative lacks the authority to make a concession, the mode collapses regardless of intent.
Why Compromise Is Not a Win/Win Strategy
Compromise has a strong reputation it hasn't earned. The textbook definition is mutual concession: each side gives up part of its goal. Both sides usually walk away feeling slightly worse, which is why the mode is often described as "lose/lose with a nicer meeting."
The practical difference shows up in a worked case. Team A wants 40 hours of shared test-environment time. Team B wants 40 hours. A compromise gives each 20. Neither can run its regression suite full length, so both parallelize and the project slips by three days anyway. Collaboration asks why each team needs the environment: A's need is a nightly regression run, B's need is a load test against production-like infrastructure. If the load test runs in a scheduled three-hour block and the regression runs in the remaining window, both get what they actually require. The units on the table were 40 hours; the interests were recoverable and weren't.
Split-rule compromise has one legitimate home: small, reversible, low-stakes decisions made under time pressure. If the choice is which of two meeting rooms to book, concede and move on.
A Step-by-Step Collaboration Protocol
The protocol below is the practical core of the article. Run the steps in this order; skipping step two is the most common failure.
1: Name the conflict as a joint problem, in writing. State it so both teams can sign the sentence. "How do we deliver the Q2 integration without either team missing its release slot" works. "Team A is blocking us" does not. 2: Separate positions from interests. Each team writes down what it is demanding and then, next to it, why. Positions are what they ask for; interests are the underlying need, fear, or commitment behind the ask. 3: Agree decision criteria before generating options. Write the 3–5 criteria both sides will use to judge any proposal—for example, total cost, schedule impact, reversibility, and who carries the residual risk. Building options before criteria invites scorekeeping. 4: Generate options without evaluation. Brainstorm alternatives first; evaluate later. Evaluated brainstorming produces only proposals the proposer already knows the other side can accept, which shrinks the solution space to compromise. 5: Select against the criteria. Score the shortlisted options against the criteria from step three. The explicit criteria are what make the outcome a joint decision rather than one side's idea that the other side tolerated. 6: Document terms, owners, and date. Write the agreement down with named owners and a review date. Undocumented collaboration decays within a sprint.
The failure-handling matters as much as the steps. If step three stalls—teams can't agree on criteria—the likely problem is that one side hasn't surfaced its real interest, often because the person in the meeting can't commit. Escalate to the sponsor with the interests and the disagreement on criteria, not with a request that the sponsor pick the winner. If a side reopens a settled point, the usual cause is an unstated constraint; ask what would have to be true for the agreement to hold.
The time cost is real and worth budgeting explicitly. A collaboration session you run yourself needs roughly 90 minutes of live call time per H2 topic above, plus 20–30 minutes per team of preparation. Using a trained facilitator adds cost; check the facilitator's current rates directly rather than working from stale figures. Compare that sum to the cost of the rescheduling, rework, and second pass that a losing team usually forces anyway.
Building Decision Criteria Both Teams Can Sign
Criteria are where good collaboration is won or lost. Four tests separate workable criteria from decorative ones: they must be measurable, they must be the same for both teams, they must be agreed before options are scored, and each side must be able to state them from memory.
A useful benchmark is the "Yes" criterion used in interest-based negotiation training—the criterion that, if met, would let you say yes to an agreement without needing to win on everything else. You and the other team each write your criterion privately, then compare. If one side's criterion contradicts the other's, that contradiction is the real conflict, and it's far easier to solve than the arguments about environment hours.
Visibility matters more than formality. If the sponsor can see the decision rubric and the options that were considered, the losing party has a harder time relitigating the outcome later. A visible rubric converts a perceived loss into a documented trade-off, which is exactly the difference between one team exiting resentful and one team exiting with a clearer picture of what it holds.
When Collaboration Fails and What to Use Instead
Collaboration is the right default, not the universal answer. Three situations in project work reliably break it: a genuine emergency, a non-negotiable of legal or safety origin, and a side that has no intention of solving the problem.
The TKI framework gives named alternatives for each. In an emergency, forcing brings the shortest path to action; an incident commander assigning roles is forcing, and it's correct there. Against a hard compliance requirement, accommodation—conceding on that point because fighting it is futile—is the honest label. When the counterparty is adversarial, competing or a formal escalation protects your interests more than another collaborative session will.
A nuance from research on team conflict: too little task conflict produces groupthink and missed errors, and too much produces paralysis. A state of low task disagreement isn't the mark of good teamwork, it's a yellow flag. Deliberate structured dissent before commitment—the pre-mortem, the red team—is how you install task conflict on purpose without letting it turn personal.
The Conflict-Mode Comparison at a Glance
The table below is the artifact to take into your next steering meeting.
| Mode | Assertiveness | Cooperativeness | Typical time cost | Use on a project when |
|---|---|---|---|---|
| Collaboration | High | High | Highest — ~90 min session + prep per issue | Both teams hold something the other needs; solution space can be expanded |
| Compromise | Medium | Medium | Low–medium | Small, reversible decision; time pressure; no clear joint gain available |
| Accommodation | Low | High | Low | Issue matters far more to the other team; preserving the relationship outweighs this win |
| Forcing | High | Low | Lowest | Emergency, safety, legal threshold, or a manager must act fast |
| Avoidance | Low | Low | Near zero, deferred | Issue is trivial, cooling is needed, or the conflict has no owner yet |
Read the time-cost column against the stakes. The cheapest mode is not automatically the right mode, and the most expensive mode is not automatically the noble one. Apply collaboration to decisions that are expensive to redo; apply forcing where delay carries the greater cost.
What a Real Collaborative Outcome Looks Like
A worked example makes the difference concrete. Two delivery teams, platform and applications, are fighting over a shared integration environment during a release window. Platform needs the environment for a version upgrade. Applications needs it for end-to-end testing. Standard position-taking produces a schedule where each team gets half the window, and both products ship with the integration untested on the new version.
A collaborative session changes the shape of the problem rather than the split. The interest behind platform's demand is that the upgrade be validated under production-like load before release. The interest behind applications' demand is that the end-to-end suite pass on the release candidate. The joint constraint is that testing must run on the post-upgrade version. That constraint actually forces a sequence—upgrade first, then test—which neither side had noticed while arguing about time slices.
The final arrangement puts the upgrade in a fixed morning block, hands the environment to applications for the rest of the day, and adds a joint smoke test at the end of the week. Both teams' success criteria get met. The option space expanded because the interests were visible.
Residual risk stays with whoever's area it sits in, and the written agreement should say so. If the upgrade destabilizes the environment mid-test, platform owns the fix and applications owns the re-run. Naming residual risk is how collaboration avoids the quiet transfer of blame that undoes agreements later.
The "Nobody Loses" Checklist Before You Commit
Run this check before anyone signs off. Skipping it is the most common reason a collaborative agreement unravels halfway through the sprint.
Can the lowest-level implementer restate the agreement? If they can't, the message hasn't traveled. Does every agreed criterion have an owner and a measurement date? Criteria without owners become opinions within two weeks. Is the residual risk assigned in writing? Undefined risk quietly becomes the losing team's problem. Is there a review date? Conditions change; the agreement needs a scheduled re-check, not a permanent status. Would either team sign again today? Ask the awkward question. A "no" here still leaves time to fix it.
The checklist is short because the failure modes are narrow. Most collaboration failures in projects trace to one of two causes: the decision was made by people lacking authority, or it was never written down. The checklist catches both jobs if you run it honestly.
Designing a Session That Both Teams Leave Winning
How you run the meeting decides the outcome. The biggest determinant isn't the seating plan or the agenda format—it's whether each team's representative has the authority to make a concession on the spot.
Prepare with a pre-read, not a slide deck. Send a one-page summary of the conflict, both teams' written positions and interests, and the draft criteria, 48 hours ahead. Packing the conflict into a room to introduce both positions cold usually yields a fight between the argued positions rather than the actual problem.
Facilitate with visible notes. The person taking notes should mirror it—two screens or a projected document—so both sides see their contribution. When it's your own frame, say so out loud: "This is my framing; correct it where it's off." That's a short opening move, and it's what keeps a facilitated conversation from drifting into an exchange of demands.
Draft the agreement line by line at the end of the session and read it back. Leave without a written record and you'll have a second fight tomorrow about what the first fight was even about.
FAQ
Is compromise ever the right answer for a win/win goal?
No—compromise produces a partial win by definition, so it's a bounded fallback rather than a win/win method. Use it for decisions that are small, reversible, and time-pressed, where the cost of a full collaboration session exceeds the value of the decision itself. For anything that touches schedule or scope of either team's release, compromise tends to reappear as rework within the same quarter.
Can you force a win/win solution from above?
Forcing a collaborative outcome fails about as often as forcing any other outcome, for two reasons. The teams didn't build the solution, so they don't own it, and they can't name their own interests—which means the agreement leaves the real disagreement untouched. A manager's legitimate play is to mandate the process and the deadline, not the answer.
What does this conflict actually cost if it stays unresolved?
Budget the process cost against rework cost. A collaboration session on a single project issue takes roughly 90 minutes plus 20–30 minutes of preparation per team. A forced resolution that a losing team resists often costs days of rework, missed test windows, and a second decision cycle. For two teams of five, one meeting hour is ten person-hours—so a 90-minute session is about 15 person-hours before preparation, and you should compare that to the reschedule cost honestly.
What happens when both teams race for the same mode?
Collaboration collapses if only one side is playing it. The party still arguing from positions takes the expanded option space available and returns nothing, and the cooperating side walks out with a worse result than compromise would have produced. If your counterpart won't disclose interests after a reasonable attempt, switch modes and escalate with the interests documented.
When should a manager choose forcing instead?
When delay is the dominant risk—an incident, a safety threshold, a legal deadline, or a launch window that closes in hours—forcing is correct and should be labeled as such to the team. After the event, do a short collision review so the decision context doesn't become the team's permanent culture.
Does collaboration count if only one side gets its outcome?
Yes, if the outcome is jointly chosen—but only if both teams' success criteria were on the table and the trade-off was explicit. That's the test to apply before calling an outcome win/win. If one side's success criteria never entered the conversation, the agreement is a disguised concession, and it tends to resurface as passive resistance on delivery.
What if the two teams report to different sponsors?
Run one collaboration session with a single decision owner from each side who has authority over that team's commitments. Two sponsors with unequal authority usually turn collaboration into forcing by proxy. When authority is split, the first move is getting the two decision owners into the same room, or getting a joint sponsor who can resolve the residual conflict if the session fails.
A Practical Frame for the Next Conflict
Collaboration earns its higher cost when the conflict is about work rather than status, when both teams hold something the other needs, and when the person in the room can concede. That's the selection rule to carry into your next project conflict: run the protocol and don't hand the resolution down from above unless emergency, legal, or adversarial conditions force you to a different mode. The record you write down is what makes the win/win survive past the meeting.