Purpose
Score a list of proposed work items on business impact versus engineering effort, sort them into a defensible priority order, and surface the low-value work that should be cut — so an engineering manager walks out of triage with a ranked list, not a debate.
When to Use
Invoke this skill when the user asks to:
- Triage a backlog and decide what to keep, defer, or kill
- Choose what goes into the next sprint from a candidate list
- Rank competing feature requests or tech-debt items
- Justify why a low-value item is being deprioritized to stakeholders
- Compare a "quick fix" against a larger investment on the same goal
- Cut a bloated backlog down to the items that actually move a metric
Inputs Required
Before starting, collect:
- Work items: The list of candidates — one line each (feature, bug, tech debt, request)
- Team goal: The outcome this team is accountable for this quarter (e.g., reduce churn, increase activation, ship platform stability)
- Time horizon: Sprint, quarter, or "next thing we pick up"
- Team capacity: Approximate engineer-weeks available, or team size
- Known dependencies or constraints: Anything that blocks or forces sequencing
- Existing estimates (optional): Any effort estimates already attached to items
If effort estimates are missing, infer a relative t-shirt size (S/M/L/XL) per item and state the assumption explicitly so the manager can correct it.
Instructions
Step 1: Define the Impact Axis Against the Team Goal
Score each item's Business Impact 1–5, anchored to the stated team goal — not generic "niceness."
- 5 — Directly moves the team's primary metric; measurable within the horizon
- 4 — Strong indirect contribution or unblocks a high-impact item
- 3 — Real value but tangential to the primary goal
- 2 — Marginal; quality-of-life or low-frequency improvement
- 1 — No clear line to the goal; "someone asked for it"
For each score, write one sentence naming the specific metric or outcome affected. If an item has no traceable line to the goal, score it 1 or 2 — do not inflate.
Step 2: Define the Effort Axis
Score each item's Engineering Effort 1–5, where higher means more costly.
- 1 — Hours; trivial, well-understood
- 2 — A day or two; low risk
- 3 — Several days to a week; some unknowns
- 4 — Multi-week; cross-component or coordination required
- 5 — Multi-sprint; high uncertainty, research, or migration
Factor in hidden effort: testing, migration, review load, and operational risk — not just keyboard time. Note any item where effort is uncertain (flag as "estimate volatile").
Step 3: Assign Each Item a Quadrant
Plot impact against effort and label the quadrant:
- Quick Win — High impact (4–5), Low effort (1–2): do first
- Big Bet — High impact (4–5), High effort (3–5): plan, sequence, resource deliberately
- Fill-In — Low impact (1–3), Low effort (1–2): batch when capacity allows; never displaces a Big Bet
- Money Pit — Low impact (1–3), High effort (3–5): cut or explicitly defer — these are the items to kill
Step 4: Compute a Priority Score and Rank
For each item, compute Priority Score = Impact ÷ Effort (round to two decimals). Rank all items descending. Use the quadrant to break ties and sanity-check the math:
- A Money Pit should never outrank a Quick Win regardless of arithmetic
- If a high-score item has a hard dependency, note that it cannot start until the blocker clears
Step 5: Build the Cut List
Explicitly list the items recommended for cutting or indefinite deferral (all Money Pits, plus any Fill-In that has aged or duplicates a higher-ranked item). For each, give a one-line reason a stakeholder will accept (e.g., "No path to the activation metric; 3 engineer-weeks better spent on Quick Win #2").
Step 6: Produce the Sprint/Horizon Recommendation
Given team capacity, recommend what fits in the stated horizon:
- Fill capacity from the ranked list top-down, respecting dependencies
- State remaining capacity after each item to show the math
- Call out the one item the manager should defend hardest if challenged, and the one most likely to get cut under pressure
Output Format
- Structure: Scoring table → quadrant grouping → cut list → horizon recommendation
- Length: 700–1,100 words depending on item count
- Format: A scored table (Item | Impact | Effort | Quadrant | Priority Score | Note), followed by short prose sections
- Sections: Scored Table → Ranked Priority Order → Cut List → Sprint/Horizon Recommendation → The One You'll Have to Defend
- Tone: Decisive engineering manager — quantified, opinionated, willing to name what to kill
Quality Checklist
Before delivering output, verify:
- [ ] Every input work item appears in the scored table exactly once
- [ ] Each impact score names the specific metric or outcome it ties to
- [ ] Each item has a quadrant label consistent with its impact/effort scores
- [ ] Priority Score = Impact ÷ Effort is computed and items are ranked descending
- [ ] No Money Pit outranks a Quick Win in the final order
- [ ] The cut list is non-empty (or it explicitly states why nothing should be cut)
- [ ] The horizon recommendation respects the stated capacity and shows remaining capacity
Notes
- The most common failure in triage is impact inflation — every stakeholder believes their item is a 5. Anchoring impact strictly to the team's one stated goal is what makes this useful.
- Effort estimates are the weakest input; treat "estimate volatile" flags as a prompt to ask one engineer before committing a Big Bet.
- Quick Wins should ship before Big Bets even when a Big Bet has higher raw impact — early momentum and capacity reclaimed from Quick Wins often fund the bet.
- If two items target the same metric, note whether they're additive or redundant; redundant lower-ranked items belong on the cut list.
How to use this skill
This file is a Claude Code skill in Anthropic's standard format — markdown with YAML frontmatter at the top.
Installing in Claude Code:
~/.claude/skills/<folder-name>/SKILL.md— auto-loads in every session anywhere on your machine (recommended for skills you'll use across multiple projects)<your-project>/.claude/skills/<folder-name>/SKILL.md— auto-loads only when working in that project (recommended for project-specific skills)
Restart Claude Code or open a new session to register.
Using in Claude.ai, ChatGPT, Gemini, or other AI tools:
- Drag this file into a new chat, or copy and paste its contents
- Send your task or request as your next message — the AI will apply the skill where it fits
For Claude Code skill details: https://docs.anthropic.com/en/docs/claude-code/skills