SCOPECREEP SURVIVAL

Caffeine is the most widely used psychoactive substance in software teams. In short bursts it genuinely improves focus, reaction time, and mood — but overconsumption creates dependency and crashes. Use it to recover momentum, not as a baseline.

"Double espresso. It doesn't fix the sprint. It makes you feel like it might."

A well-written, clearly scoped ticket is genuinely rare and genuinely powerful. It removes ambiguity from the engineer's path, prevents back-and-forth, and accelerates delivery. The INVEST criteria (Independent, Negotiable, Valuable, Estimable, Small, Testable) is the gold standard for writing tickets that actually get done.

"Estimated: 1 point. Actual: unknown. Status: In Progress, since Q1."

Rubber duck debugging is the practice of explaining your problem out loud — to a rubber duck, a colleague, or anyone who will listen. The act of articulating the problem forces you to examine your own assumptions. It is one of the most effective debugging and problem-solving techniques in software, and transfers directly to PM decision-making. Use this when stuck, not just when broken.

"The duck has heard every architectural debate since 2019. It is unimpressed but supportive."

A product roadmap communicates direction, priorities, and timelines to stakeholders. It creates alignment — temporarily. Good roadmaps are outcome-focused (not feature-lists), clearly prioritised, and explicitly marked as subject to change. In this game, pulling out the roadmap buys you trust and rallies the team around a shared goal.

"Subject to change. Has always been subject to change. Will always be subject to change."

Prompt engineering is the discipline of designing inputs to LLMs to reliably produce useful, safe outputs. A validated prompt template represents hours of iteration, red-teaming, and edge-case handling. Using one improves model quality immediately and reduces risk because it's been safety-reviewed — unlike ad-hoc prompts that drift in production.

"Engineered through 400 iterations. Handles edge cases. Mostly."

Key Performance Indicators quantify product success in terms stakeholders understand. A live KPI dashboard silences opinion-based debates by replacing "I think" with "here is what the data shows." For AI products, KPIs should include both business metrics (conversion, retention) and model metrics (accuracy, latency, hallucination rate).

"Live data. Unless the tracking is broken. Which it was, last Tuesday."

A Data Processing Agreement (DPA) is a legally binding contract required under GDPR and similar frameworks when personal data is processed by a third party. For AI products that process user data, a signed DPA is not optional — it's a legal obligation. Using it in the game reflects the real practice of proactively addressing data governance before regulators come asking.

"87 pages. You have read 4 of them. Legal has read all 87. They are not the same 4."

Proactive stakeholder communication is one of the highest-leverage PM skills. A well-crafted async update — status, blockers, next steps — preempts escalation, builds credibility, and respects people's calendars. In AI projects where uncertainty is high, regular written updates keep stakeholders informed without pulling engineers into unnecessary syncs.

"Subject: Quick Update. Body: 400 words. Subtext: Please don't escalate."

OKRs (Objectives and Key Results) create alignment between company strategy and team execution. A well-written OKR gives the team a clear goal (Objective) and measurable outcomes to track progress (Key Results). In this game, publishing your OKRs reassures stakeholders and focuses the team — both of which translate directly to trust and velocity.

"Key Results are measurable this quarter. That's already an improvement."

Meeting overload is one of the top contributors to developer burnout and lost focus time. Research shows that developers need uninterrupted blocks of 2–4 hours to do deep work — fragmented schedules destroy this. Cancelling a meeting gives your team focus time back, at the cost of a small hit to stakeholder relations. Use it when the team is overwhelmed, not when you just don't feel like attending.

"This meeting could have been an email. It will now be an email."

Technical debt is the implied cost of future rework caused by choosing a fast, expedient solution now over a better but slower one. Like financial debt, small amounts are manageable and sometimes strategic. But letting it compound — by intentionally accruing it without a repayment plan — creates a "bomb" that detonates unpredictably. Martin Fowler's Technical Debt Quadrant is essential reading for any PM who ships AI products under time pressure.

"You'll deal with it next sprint. Next sprint is a myth."

Workplace stress in tech is chronic and cumulative. The APA identifies role ambiguity, lack of control, and workload as the top drivers — all things a PM directly influences. Physical stress-relief tools are a symptom of a broken system, not a solution. In the game the stress ball represents the PM taking a moment to acknowledge the team's pressure and proactively managing morale before it collapses.

"Dented on one side from the Q3 scope creep incident. A survivor."

A Model Card is a short document that describes what an AI model does, what it was trained on, its known limitations, evaluation results, and intended use cases. Introduced by Google in 2019, they've become an industry standard for responsible AI disclosure. Publishing a model card simultaneously improves perceived quality (by demonstrating rigour) and reduces compliance risk (by showing regulators you've documented your model's behaviour).

"The document that proves you thought about it before shipping."

Crunch — forcing a team to work unsustainable hours to hit a deadline — delivers short-term velocity at the cost of quality, morale, and retention. Research consistently shows that hours beyond 50/week produce negative returns in knowledge work. Energy drinks in the game represent that same tradeoff: you get a burst of speed, but you're burning your team's health to do it. Use only in genuine emergencies.

"Third one this week. The engineers can tell. You can tell they can tell."

The sprint retrospective is Agile's built-in learning loop — a structured team conversation about what went well, what didn't, and what to change next sprint. When done well (psychologically safe, action-item focused, followed up), retros meaningfully improve both team health and throughput. The Atlassian guide below covers the Start/Stop/Continue format that works best for most teams.

"Action items: 4. Resolved action items: 1. Feelings processed: many. Net positive."