startup-pivoting

What this skill does

# Startup Pivoting

## Scope

**Covers**
- Deciding whether to **pivot vs persevere** when a product/startup is stuck (typically pre- or early-PMF)
- Turning a pivot debate into an executable **Pivot Decision & Execution Pack** (not vibes)
- Designing a **time-boxed pivot validation + execution plan** with clear decision gates

**When to use**
- “Should we pivot?”
- “We’re stuck pre-PMF / growth has stalled.”
- “We think the ICP/market is wrong—help us change direction.”
- “We need a pivot options map and a concrete plan to validate one quickly.”
- “Make a pivot decision memo we can share with the team/investors.”

**When NOT to use**
- You don’t have a product or real customer evidence yet (do discovery/problem framing first, e.g., `problem-definition`)
- You only need **incremental optimization** (pricing tests, onboarding tweaks, activation/retention work) and direction is not in question
- You’re choosing between many roadmap bets within an agreed strategy (use `prioritizing-roadmap`)
- You want the agent to “pick a new startup idea” from scratch (use `startup-ideation`)

**Human checkpoint (required)**
- A pivot is a high-stakes strategic decision. This skill produces decision-ready artifacts and a plan, but a human owner must make the final call.

## Inputs

**Minimum required**
- What you sell/build today (product summary + current target customer)
- The “stuck” symptoms + evidence (metrics, user feedback, pipeline, retention/churn, qualitative signals)
- Runway/timebox (months of runway or a decision deadline)
- Constraints/non-negotiables (compliance, brand/trust, margins, platform, team capabilities)
- Current “theory of winning” (who + problem + why you win)

**Missing-info strategy**
- Ask up to 5 questions from [references/INTAKE.md](references/INTAKE.md).
- If answers aren’t available, proceed with explicit assumptions and offer 2 scope options (lean 60–90 min analysis vs thorough 1–2 day pack).

## Outputs (deliverables)

Produce a **Pivot Decision & Execution Pack** in Markdown (in-chat; or as files if requested):

1) **Context snapshot** (what’s true today; constraints; runway; decision owner)
2) **Stuck diagnosis** (symptoms → likely causes → evidence gaps)
3) **Exhaustion check** (“have we exhausted the possibilities?”) + the last best non-pivot moves (time-boxed)
4) **Pivot options map** (4P pivot grid + 10% vs 200% classification)
5) **Chosen pivot thesis** (who/problem/promise) + **success metrics** + **kill criteria**
6) **Validation plan** (customer learning + experiments; decision gates; what would change your mind)
7) **Execution plan** (pivot sprint plan; cut list; resourcing; comms; risks)
8) **Risks / Open questions / Next steps** (always included)

Templates: [references/TEMPLATES.md](references/TEMPLATES.md)  
Expanded guidance: [references/WORKFLOW.md](references/WORKFLOW.md)

## Workflow (7 steps)

### 1) Frame the decision (and the clock)
- **Inputs:** Request + runway/timebox + [references/INTAKE.md](references/INTAKE.md).
- **Actions:** Define the decision: **pivot vs persevere vs shut down** (or “pivot A vs pivot B”). Name the decision owner and the decision date. Capture non-negotiables.
- **Outputs:** Context snapshot + decision statement.
- **Checks:** The decision is binary/explicit and time-bounded (no “let’s think about it”).

### 2) Diagnose what’s actually “stuck”
- **Inputs:** Metrics, funnel, retention/churn, pipeline, user feedback, win/loss notes.
- **Actions:** Summarize symptoms and hypothesize causes. Separate **signal** (real demand/value issues) from **execution** (distribution/onboarding/pricing) issues.
- **Outputs:** Stuck diagnosis + evidence inventory + top evidence gaps.
- **Checks:** You can state the top 1–3 bottlenecks and what data would falsify them.

### 3) Run the exhaustion check (Butterfield rule)
- **Inputs:** Stuck diagnosis + constraints + prior attempts.
- **Actions:** Use [references/TEMPLATES.md](references/TEMPLATES.md) to complete the **Exhaustion Check**: list the most credible non-pivot levers and whether they were tried well. Identify the “last best” 1–3 non-pivot moves worth time-boxing (if any).
- **Outputs:** Exhaustion Check + time-boxed “last best tries” plan (or explicit rationale for skipping).
- **Checks:** If you recommend pivoting, you can explain why remaining non-pivot levers are unlikely/too slow vs runway.

### 4) Generate pivot options (require at least one 200% pivot)
- **Inputs:** Theory of winning + evidence + constraints.
- **Actions:** Create 4–8 options using a **4P pivot grid** (Problem, Persona, Product, Positioning/Package). Classify each as ~**10%** (small tweak) vs **200%** (meaningfully different bet). Include at least one 200% option.
- **Outputs:** Pivot options map with a short “why this could win” and “what would have to be true”.
- **Checks:** Each option is distinct, falsifiable, and has a plausible path to distribution.

### 5) Select a pivot thesis + metrics + kill criteria
- **Inputs:** Options map + decision criteria (runway, strengths, market, moat).
- **Actions:** Pick the best option (or top 2). Write a **Pivot Thesis Card** and define success metrics (North Star + 2–5 leading indicators) plus guardrails. Define kill criteria and a decision gate date.
- **Outputs:** Pivot thesis + metrics + kill criteria.
- **Checks:** Metrics are computable; kill criteria are real (not “keep going until it works”).

### 6) Build the validation + execution plan (Todd Jackson rule)
- **Inputs:** Pivot thesis + runway + team capacity.
- **Actions:** Create a time-boxed **pivot sprint**: customer learning plan, experiments, build scope, and what to stop building. Include a comms plan (team/investors/customers) and a rollback/exit plan if results fail.
- **Outputs:** Validation plan + execution plan (owners, timeline, decision gates).
- **Checks:** Plan fits runway; includes a cut list; includes at least one “hard truth” test that could disconfirm the thesis.

### 7) Quality gate + finalize pack
- **Inputs:** Full draft pack.
- **Actions:** Run [references/CHECKLISTS.md](references/CHECKLISTS.md) and score with [references/RUBRIC.md](references/RUBRIC.md). Add **Risks / Open questions / Next steps** and confirm the human checkpoint (decision owner/date).
- **Outputs:** Final Pivot Decision & Execution Pack.
- **Checks:** A stakeholder can read it async and understand: (a) why you’re pivoting, (b) what you’ll do, (c) how you’ll know, (d) when you’ll decide.

## Quality gate (required)
- Use [references/CHECKLISTS.md](references/CHECKLISTS.md) and [references/RUBRIC.md](references/RUBRIC.md).
- Always include: **Risks**, **Open questions**, **Next steps**.

## Examples

**Example 1 (B2B SaaS stuck pre-PMF):** “We built an AI support copilot for SMBs. Trials convert, but retention is poor and sales cycles are long. We have 6 months of runway. Should we pivot, and if so how?”  
Expected: a Pivot Decision & Execution Pack with an exhaustion check (pricing/onboarding/ICP), 4–8 pivot options including at least one 200% pivot, a chosen thesis with metrics/kill criteria, and a 4–6 week pivot sprint plan.

**Example 2 (Consumer plateau):** “Our language learning app growth stalled and D30 retention is low. We suspect our promise is wrong. Create a pivot options map and a validation plan.”  
Expected: a 4P pivot grid that includes positioning/package changes and at least one new persona/problem angle, plus a time-boxed validation plan with decision gates.

**Boundary example:** “Just tell us what to pivot to—no metrics, no customers, no constraints.”  
Response: explain that pivoting without evidence is guesswork; ask 3–5 intake questions, propose a discovery/validation sprint, and only then produce a pivot thesis and plan.