Mission
By the end of this chapter, you can turn cross-functional disagreement into an evidence-based negotiation, route unresolved authority correctly, and close a decision without erasing dissent.
- Measurable outcome: Facilitate a material disagreement from positions to interests, options, criteria, decision, commitment, and revisit triggers.
- Prerequisites: Chapters 16, 21, and 22; one current disagreement or the Helios case.
- Work product: A conflict map, negotiation brief, and closure record.
- Time: 80–100 minutes.
Before you read: Predict → Commit → Connect
Predict: Product says “ship autonomy now”; Security says “no autonomous actions.” What question could reveal compatible interests beneath those positions?
Commit: Write one disagreement as two neutral claims without naming either party as the problem.
Connect: Recall a meeting where everyone “aligned,” then behaved differently. Which decision, owner, or commitment was never explicit?
Diagnose the conflict before solving it
Conflict is not one thing. Task conflict concerns what should be done. Process conflict concerns how work, authority, or resources are arranged. Relationship conflict concerns identity, disrespect, trust, or interpersonal history. Real disagreements can contain all three, and these labels are prompts rather than clinical diagnoses.
Task disagreement can improve the evidence considered when people can challenge ideas without personal attack and the organization can still decide. Persistent process ambiguity or relationship harm can overwhelm that benefit. Do not celebrate conflict in the abstract; shape the conditions under which relevant disagreement becomes useful.
Begin with a neutral conflict statement:
The program needs a first-release tool-action policy that delivers measurable customer value while meeting authorization, privacy, audit, and operational constraints. Product and Security currently recommend different exposure levels.
This frames a shared decision rather than a contest between personalities.
Some conduct or employment matters require manager, HR, legal, ethics, or safety channels. A TPM should not treat every concern as a facilitation exercise.
Move from positions to interests and criteria
A position is a proposed answer: “human approval for every refund.” An interest is the need beneath it: preventing unauthorized financial action, preserving auditability, protecting customers, learning quickly, or reducing handling time. Interests can still conflict, but they create more design space.
Principled negotiation practices emphasize interests, options, and objective criteria. In a program, use:
- Shared decision question: What exactly must be decided?
- Interests and constraints: What outcome, obligation, risk, or identity matters to each party?
- Evidence: What is known, disputed, or missing?
- Options: Include hybrids, staged exposure, experiments, and the no-agreement consequence.
- Criteria: What standards will compare options?
- Authority and timing: Who decides, and by when?
“Objective criteria” does not mean value-free. Policies, reliability targets, cost thresholds, customer research, and evaluation results embed choices. Agree on relevant criteria and their owners rather than using numbers as weapons.
Generate options before advocating one. A staged experiment can convert an argument about predictions into evidence, but only if exposure is ethically and operationally acceptable. Do not experiment with unapproved material harm.
Close the decision and commitment
Debate without closure creates meeting debt. Closure requires:
- an explicit decision by the authorized owner;
- rationale, constraints, and material dissent;
- actions with owners and dates;
- who must commit to execution;
- what remains unresolved;
- evidence and triggers for revisit;
- a communication path to affected people.
Disagree and commit is not a command to stop raising safety, legal, ethical, or newly evidenced concerns. It means that after a legitimate decision within authority, participants execute it while preserving defined escalation and revisit paths. If authority was wrong, evidence was concealed, or a control was breached, the decision system must reopen.
Avoid false compromise. Splitting the difference between safe and unsafe is not inherently wise. The decision must meet non-negotiable controls and explicitly trade dimensions that are actually tradable.
Recurring case: Helios Support
Product wants Helios to issue low-value refunds automatically because recommendations alone do not change resolution time enough. Security wants human approval for all write actions because the tool pathway has not been validated against credential abuse. Operations worries that an extra approval queue will erase the customer benefit.
The TPM identifies shared interests: correct and fast resolution, bounded loss, auditability, and learning before broader exposure. The parties agree on criteria: authorization integrity, policy conformance, measurable customer improvement, operator recoverability, and monitoring cost.
They produce three options: recommendations only; capped autonomy for a narrow authenticated intent with confirmation, monitoring, audit, and kill switch; or broad autonomy. A payments risk owner approves the capped pilot. Security records residual concerns about abuse detection; Product commits to the cap; Operations owns the runbook. The decision reopens if an unauthorized action occurs, audit quality falls, or the evidence window completes.
Decision rights: Who owns what?
- Participants: own accurate evidence, respectful conduct, and faithful representation of interests.
- Domain owners: own technical, product, operational, or control recommendations in their authority.
- Authorized decider: owns the final trade-off and residual exposure within delegated bounds.
- People managers or HR: own conduct and employment matters that exceed program facilitation.
- TPM: frames the decision, makes interests and evidence comparable, facilitates options, records dissent, and drives closure. The TPM does not coerce agreement or invent authority.
I do
I meet parties separately only to understand context, not to make secret commitments. I restate each view in language the other side accepts. I write the shared decision question and ask each group for interests, evidence, non-negotiable constraints, and one option other than its preferred position.
In the joint session, I confirm criteria before comparing options. At the end, I ask the decider to state the decision aloud, then ask each action owner to state their commitment and concerns. I record both.
We do
Together, repair “Engineering is being difficult and refuses the deadline.”
We replace it: “Product needs the campaign capability by September 1. Engineering's current forecast is September 20 because the proposed design lacks a safe replay path. We need to decide among reduced scope with replay, additional capacity if it changes the constraint, or a later campaign. Reliability acceptance remains with the service owner; the sponsor owns the campaign trade-off.”
You do
Choose a live disagreement. Classify task, process, and relationship elements. Write a neutral shared question, each party's positions and interests, known/disputed evidence, at least three viable options, and decision criteria. Identify the authorized decider and deadline. Facilitate or simulate closure, including dissent and revisit triggers.
Show the model answer
Model answer and 0–4 rubric
Question: Under what conditions may Helios execute a refund without human approval during the first production evidence window? Interests: Product: faster correct resolution and learning; Security: prevent unauthorized financial actions; Operations: recoverability and sustainable review load; Payments Risk: bounded exposure and auditability. Criteria: authenticated intent, authorization integrity, policy conformance, maximum exposure, customer benefit, audit coverage, kill-switch recovery, and operational cost. Decision: Narrow capped pilot with explicit eligible intents, user confirmation, sampled audit, anomaly monitoring, and kill switch. Dissent: Security believes abuse-detection coverage remains weak; concern is recorded and becomes a revisit trigger. Commitment: All groups execute the approved pilot; any unauthorized action invokes immediate stop and incident handling.
Rubric
- 0 (Missing): Personal blame, repeated positions, or no authorized decision.
- 1 (Emerging): A facilitated discussion occurs, but interests, criteria, or closure are vague.
- 2 (Functional): Options and a decider exist; dissent, commitments, or revisit triggers are incomplete.
- 3 (Strong): Neutral framing, interests, evidence, criteria, authority, closure, and recorded dissent are explicit.
- 4 (Decision-ready): Level 3 plus ethical experimental bounds, verified downstream commitment, conduct routing, and proof that the decision stayed closed until legitimate new evidence appeared.
Pause & Recall
Without looking, distinguish task, process, and relationship conflict. Name the six moves from shared question to authority. Connect to Chapter 22: what conditions let someone dissent without weakening accountability?
Production lens
Track recurring conflicts by system cause: unclear interfaces, competing metrics, split authority, scarce resources, or broken commitments. Do not store speculative personality labels. Watch whether the same decision repeatedly reopens; that may reveal missing evidence, ambiguous authority, unrepresented stakeholders, or unowned actions. Route harassment, discrimination, retaliation, safety, and other protected concerns through appropriate formal channels rather than ordinary negotiation.
Workplace artifact: Conflict-to-closure brief
Neutral decision question:
Task / process / relationship elements:
Parties and formal roles:
Positions:
Interests and constraints:
Known / disputed / missing evidence:
Options, including no agreement:
Decision criteria and owners:
Decider / deadline:
Decision / rationale:
Material dissent:
Actions / commitments:
Revisit and escalation triggers:Chapter compression
Diagnose what people disagree about, frame a shared decision, move from positions to interests, compare multiple options with agreed criteria, and close through legitimate authority. Preserve dissent and reopening conditions without allowing endless renegotiation.
Retrieval deck
- Q: What distinguishes a position from an interest? A: A position is a proposed answer; an interest is the outcome, obligation, risk, or need beneath it.
- Q: Why classify conflict? A: Task evidence, process ambiguity, and relationship or conduct problems require different responses.
- Q: What makes criteria “objective” enough to help? A: They are relevant, explicit, consistently applied, and owned, even though they may still embody value choices.
- Q: What does decision closure require? A: Authorized choice, rationale, dissent, actions, commitments, communication, and revisit triggers.
- Q: When should disagree-and-commit not silence a concern? A: When new material evidence, control breaches, concealed facts, or invalid authority arise.
Spaced review
- Now: Rewrite one position as three possible interests.
- +1 day: Recreate the conflict-to-closure funnel without notes.
- +3 days: Draft three options and explicit criteria for a live disagreement.
- +7 days: Inspect a reopened decision and identify the missing closure element.
- +14 days: Test whether a closed decision remains stable until a legitimate trigger occurs.