Lesson 4 of 8 · 48 min

Conflict, disagreement, and working with hard people

Conflict skeleton, Disagree and Commit for real, eng-vs-PM and overruled worked stories, difficult-teammate framing, weak vs strong endings, and probe packs.

Conflict stories are integrity tests

Interviewers ask about disagreement to see if you can separate ideas from people, use evidence, escalate cleanly, and commit after a decision. Many builders either claim they never conflict or tell a story where they 'won' by force. Both fail. This lesson builds conflict stories that score Disagree and Commit, Earn Trust, and collaboration.

Prompts: tell me about a disagreement with a coworker; a time you pushed back on a manager/PM; a difficult teammate; when you were overruled. Same skeleton: shared goal, different bets, evidence, decision rights, commit behavior after.

Hard truth: if your only conflict stories end with the other person quitting or HR, you need different stories. Loops look for professional disagreement under shared goals, not workplace drama.

The conflict skeleton

code

1CONFLICT SKELETON23  1. Shared goal          (what you both wanted for the user/company)4  2. Different bets       (option A vs B — technical or product)5  3. Your evidence        (data, risk, constraint — not vibes)6  4. Their incentive      (what they optimized for — steelman)7  5. Decision rights      (who decides; how you escalated if needed)8  6. Outcome              (what shipped; your commit behavior)9  7. Relationship residue (still trusted? process change?)1011  Kill: villain narrative, mind-reading motives, 'I was right they were dumb'.

Steelmanning the other side is a senior signal. 'PM optimized for demo date; I optimized for mobile breakages' scores higher than 'PM did not understand engineering.'

Disagree and Commit (doing it for real)

Disagree: bring evidence, propose an alternative, escalate if blast radius requires it. Commit: once the decision owner decides, execute as if it were your idea — no quiet sabotage, no 'I told you so' theater if it fails (raise risks early instead).

code

1DISAGREE AND COMMIT — SPOKEN PATTERN23  Disagree:4    "I recommended X because [evidence]. Risk of Y is [blast radius].5     Alternative Z costs [time] and reduces [risk]."6  If overruled:7    "Understood — we'll go with Y. I'll own [execution] and watch [metric].8     If [kill criteria], I'll raise it by [time]."9  After:10    Report outcomes without score-settling. Suggest process fixes if needed.

Worked story: eng vs PM on scope

code

1PROMPT: Disagreement with a product manager.23  Shared goal: ship search improvements before a marketing launch.4  Bets: PM wanted full UI redesign in 3 weeks. I owned suggest API; redesign5        forced three mobile client breaks.6  Evidence: crash analytics + release train dates; estimated 3-week client lag.7  Their incentive: coherent demo narrative for launch day.8  Move: proposed single-flight latency fix + light UI polish; full redesign post-launch.9  Decision: TL backed the split plan; PM agreed after seeing crash risk.10  Result: p95 latency win shipped pre-launch; redesign one sprint later; relationship OK.11  Residue: launch checklist now includes client compatibility gate.

Notice: no villain. Both incentives named. Evidence before emotion. Decision rights clear. Residue improves the system.

Worked story: you were overruled

code

1PROMPT: Time you were wrong or overruled.23  I wanted a full rewrite of a flaky worker. EM ordered patch + metrics first.4  I disagreed with a costed rewrite proposal; EM held the line on time-to-mitigate.5  I committed: shipped patch in 2 days, added freshness metric, reduced pages 70%.6  Rewrite later happened with data. Learning: mitigate user harm before architecture7  purity when pages are firing.89  SIGNAL: coachability + still had a real disagree moment + clean commit.

Keep at least one story where you did not win. Interviewers plant this probe. All-win banks look curated and fragile.

Difficult people without character assassination

Frame 'difficult' as misaligned incentives, unclear contracts, or stress under load — not personality defects. Your actions: clarified ownership, wrote interfaces, brought a third party for decision rights, reduced async ambiguity.

code

1DIFFICULT TEAMMATE — SAFE MOVES23  Do:  name behaviors and impact (missed Handoffs, blocked deploys)4  Do:  show what you changed in process (written contracts, SLAs)5  Do:  involve manager when decision rights were unclear6  Don't: diagnose their personality or mental state7  Don't: claim you 'fixed them'8  Don't: end on humiliation or HR as your victory

Manager pushback

Pushing back on a manager requires extra care: respect decision rights, bring options not ultimatums, and show you can escalate on principle (safety, integrity, major risk) without being oppositional on taste.

01Bring two options with costs, not a single refusal.
02Separate safety/integrity issues from preference issues.
03Ask 'what would change your mind?' — exposes decision criteria.
04Document the decision and kill criteria after commit.
05Never threaten to quit as a negotiation move in a story.
06If you escalated above your manager, say why blast radius required it.

Weak vs strong conflict endings

code

1WEAK ENDINGS2  "We agreed to disagree."3  "They eventually saw I was right."4  "I escalated and they were moved off the project."56STRONG ENDINGS7  "Decision owner chose B; I executed B; metric hit; I still flag A for later."8  "We split the launch: my risk gate + their demo narrative both survived."9  "I was overruled; commit was clean; data later justified a phase-2 rewrite."

Probe pack for conflict stories

code

1PROBES TO PREP23  - What was their best argument?4  - What did you change about your view?5  - How did you keep the relationship usable?6  - Would you disagree the same way today?7  - What process changed so this conflict is cheaper next time?

If you cannot steelman their best argument in one sentence, deepen the story before onsite.

Company flavor

Amazon: Disagree and Commit is explicit — practice the phrase through behavior, not labels. Google/Meta: collaboration + leadership under conflict. Startups: speed of resolution and low ego. Fintech: risk and compliance disagreements need extra precision.

Conflict skill is not winning. It is preserving truth and trust long enough to ship the right thing — or commit hard to the decided thing.

articleAmazon LP — Disagree and Commit / Earn TrustAmazon Jobs articleTIH — conflict and teamwork questionsTech Interview Handbook articleExponent — behavioral conflict promptsExponent articleHello Interview — behavioralHello Interview

Checkpoint

In a conflict story, what is the best way to describe the other person's position?

AExplain why they were political / incompetent so the interviewer sides with youBSteelman their incentive and bet in one line, then contrast with your evidence-based betCSkip their view entirely and focus only on what you did

Checkpoint

You were overruled by your EM. Which ending scores highest?

AI quietly slowed the work because I knew my approach was betterBI executed the decision cleanly, watched the agreed metric, and later used data for a phase-2 when appropriateCI told the team the EM was wrong so we would be ready when it failed

Checkpoint

Weakest conflict story ending?

ADecision owner chose the demo plan; I owned the risk gate metric and raised kill criteria in writingBWe split scope so both launch narrative and client safety survived; checklist updatedCWe agreed to disagree and moved on without a decision

Checkpoint

Best first move when a peer repeatedly breaks handoffs under a launch?

AWrite a public thread diagnosing their personality and work ethicBName the behavior and impact, propose a written interface/SLA, involve a decision owner if neededCDo their handoffs silently forever so the launch succeeds

Checkpoint

Why keep at least one story where you did not 'win' the disagreement?

ABecause interviewers only hire people who lose oftenBAll-win banks look curated; probes about being overruled need a real commit storyCAmazon forbids stories where you were right

Can you tell a conflict story with steelman, evidence, decision rights, and clean commit — including one where you lost?

New to itGetting thereConfident

Conflict locked

Conflict skeleton: shared goal, bets, evidence, incentives, decision rights, commit, residue.
Disagree for real, then commit cleanly — silence is not disagreement.
Steelman; no villains; keep a story where you lost.
Difficult people → behaviors, impact, process — not character assassination.
Strong endings decide and execute; 'agree to disagree' is empty.

Next: failure, incidents, and learning without theater.

Sources

Free to read · better with Enzo

Learn it with Enzo

Save your progress, answer the checkpoints, and let Enzo quiz you on what you just read.

Continue with Enzo Sign up free