Business Analyst
Convert exploration into decisions. Three modes, picked by what the caller needs; write-idea uses the first two, executive deliverables use the third.
Mode 1 — Feature scoring (for idea/PRD discovery)
Score a feature idea so its priority is an argument, not a feeling:
Six-criteria assessment — each scored Must do / Strong / Maybe / Pass, each score justified in one line citing research or codebase findings:
| Criteria |
Question |
| Impact |
How much more valuable does this make the product? |
| Reach |
What % of users would this affect? |
| Frequency |
How often would users encounter this value? |
| Differentiation |
Does this set us apart or just match competitors? |
| Defensibility |
Is this easy to copy or does it compound over time? |
| Feasibility |
Can we actually build this? |
Leverage type — Quick Win (small effort, disproportionate value), Strategic Bet (larger effort, potentially transformative), or Compounding (gets more valuable over time: data effects, habit formation, network effects).
KPIs — 3 to 6, each with a numeric target ("> 30%", "< 200ms", "-80%") and a concrete, implementable measurement method. A KPI nobody can measure is a wish.
Viability verdict — one paragraph grounded in the research: proceed, reshape, or pass, and what would change the answer.
Mode 2 — Decision support (structure an open question)
When a discovery conversation hits an open decision, convert it into a decision-ready block instead of prose:
Which retention lever should V1 optimize for?
A) Weekly digest email (Recommended) ← suggested: cheapest to ship, measurable in one cycle
B) In-app streak mechanics
C) Usage-based notifications
Assumptions to confirm: users check email weekly; digest infra exists.
Rules: 2–3 options per decision; the suggestion always carries a one-line rationale; assumptions that would invalidate the suggestion are listed, not hidden. One decision per block — batching decisions produces rubber-stamping. The structured question tool supplies the custom answer itself; without such a tool, show the same single question in chat and accept a custom answer.
Mode 3 — Executive deliverables
For full quantitative work (KPI frameworks, dashboards, forecasts, cohort/LTV-CAC models, A/B readouts, market sizing), follow the Output Contract:
Operating rules. Every deliverable names a decision, not a topic. Every number has a source, formula, or assumption-table row. State the worst case explicitly (CI lower bound, downside sensitivity, stressed LTV). Flag data-quality gaps before conclusions — never hedge with "TBD". If the contract cannot be met, say so in the opening paragraph.
Structure. Open with ## Business Objective & Success Criteria (the decision this enables + 3–5 numeric success criteria) and close with ## Recommendations (3–7 numbered items, each with Decision / Owner (a named role, never "the team") / By when / Expected impact). Between them, use executive headings — never ## Context/## Analysis/## Summary:
- Action titles stating the insight in ≤15 words:
## Enterprise cohort NRR fell 8pts; ship usage-based pricing in Q3.
- SCR for decision memos:
## Situation → ## Complication → ## Resolution (the resolution is 60–70% of the content).
- Standard vocabulary where the type calls for it:
## Input Metrics / ## Output Metrics (KPI frameworks), ## Decision Required (before Recommendations on approval memos), ## So What (page-level takeaway), ## Assumptions & Sensitivities, ## Reconciliation (market sizing: where top-down meets bottom-up).
Anti-patterns
- Scoring without justification lines — a bare "Strong" transfers no information.
- Presenting one option with a question mark instead of genuine alternatives.
- KPI targets without measurement methods, or "improve engagement" as a KPI.
- Generic persona prose ("as an expert analyst I…") — this skill is its outputs, not its résumé.
1---2name: business-analyst3description: "Turns exploration into decision-ready analysis: scores feature ideas on six viability criteria with KPI targets, converts open questions into A/B/C options with a recommended default, and produces executive-grade deliverables (KPI frameworks, forecasts, cohort/LTV-CAC models, A/B readouts, market sizing). Use during idea/PRD discovery to score or structure a decision, and whenever the task is quantitative decision support. Don't use for software implementation, UI work, or database migration."4metadata:5 category: discovery6 tags: [product, research, requirements]7 version: 0.0.38 author: Marcio Altoé9 source: https://skillmd.com/skills/marcioaltoe/business-analyst10 upstream: https://github.com/marcioaltoe/skills11version: 0.0.312---1314# Business Analyst1516Convert exploration into decisions. Three modes, picked by what the caller needs; `write-idea` uses the first two, executive deliverables use the third.1718## Mode 1 — Feature scoring (for idea/PRD discovery)1920Score a feature idea so its priority is an argument, not a feeling:21221. **Six-criteria assessment** — each scored `Must do / Strong / Maybe / Pass`, each score justified in one line citing research or codebase findings:2324 | Criteria | Question |25 | ------------------- | --------------------------------------------------- |26 | **Impact** | How much more valuable does this make the product? |27 | **Reach** | What % of users would this affect? |28 | **Frequency** | How often would users encounter this value? |29 | **Differentiation** | Does this set us apart or just match competitors? |30 | **Defensibility** | Is this easy to copy or does it compound over time? |31 | **Feasibility** | Can we actually build this? |32332. **Leverage type** — Quick Win (small effort, disproportionate value), Strategic Bet (larger effort, potentially transformative), or Compounding (gets more valuable over time: data effects, habit formation, network effects).343. **KPIs** — 3 to 6, each with a numeric target ("> 30%", "< 200ms", "-80%") and a concrete, implementable measurement method. A KPI nobody can measure is a wish.354. **Viability verdict** — one paragraph grounded in the research: proceed, reshape, or pass, and what would change the answer.3637## Mode 2 — Decision support (structure an open question)3839When a discovery conversation hits an open decision, convert it into a decision-ready block instead of prose:4041```text42Which retention lever should V1 optimize for?4344A) Weekly digest email (Recommended) ← suggested: cheapest to ship, measurable in one cycle45B) In-app streak mechanics46C) Usage-based notifications4748Assumptions to confirm: users check email weekly; digest infra exists.49```5051Rules: 2–3 options per decision; the suggestion always carries a one-line rationale; assumptions that would invalidate the suggestion are listed, not hidden. One decision per block — batching decisions produces rubber-stamping. The structured question tool supplies the custom answer itself; without such a tool, show the same single question in chat and accept a custom answer.5253## Mode 3 — Executive deliverables5455For full quantitative work (KPI frameworks, dashboards, forecasts, cohort/LTV-CAC models, A/B readouts, market sizing), follow the Output Contract:5657**Operating rules.** Every deliverable names a decision, not a topic. Every number has a source, formula, or assumption-table row. State the worst case explicitly (CI lower bound, downside sensitivity, stressed LTV). Flag data-quality gaps before conclusions — never hedge with "TBD". If the contract cannot be met, say so in the opening paragraph.5859**Structure.** Open with `## Business Objective & Success Criteria` (the decision this enables + 3–5 numeric success criteria) and close with `## Recommendations` (3–7 numbered items, each with **Decision / Owner** (a named role, never "the team") **/ By when / Expected impact**). Between them, use executive headings — never `## Context`/`## Analysis`/`## Summary`:6061- **Action titles** stating the insight in ≤15 words: `## Enterprise cohort NRR fell 8pts; ship usage-based pricing in Q3`.62- **SCR** for decision memos: `## Situation` → `## Complication` → `## Resolution` (the resolution is 60–70% of the content).63- Standard vocabulary where the type calls for it: `## Input Metrics` / `## Output Metrics` (KPI frameworks), `## Decision Required` (before Recommendations on approval memos), `## So What` (page-level takeaway), `## Assumptions & Sensitivities`, `## Reconciliation` (market sizing: where top-down meets bottom-up).6465## Anti-patterns6667- Scoring without justification lines — a bare "Strong" transfers no information.68- Presenting one option with a question mark instead of genuine alternatives.69- KPI targets without measurement methods, or "improve engagement" as a KPI.70- Generic persona prose ("as an expert analyst I…") — this skill is its outputs, not its résumé.