continuous-discovery

name	continuous-discovery
description	Build a continuous product discovery practice that generates a steady stream of customer insights. Outputs weekly interview cadence, opportunity identification, assumption testing, and discovery-to-delivery connection.
argument-hint	["team size","product stage","existing discovery practices","customer access"]
allowed-tools	Read, Write

Continuous Discovery

Continuous discovery (from Teresa Torres) is the practice of making regular, small customer touchpoints a team habit — rather than infrequent large research projects. The goal is to continuously generate customer insights that drive better product decisions, rather than relying on big discovery sprints that quickly become stale.

The Continuous Discovery Habit

## Core Habit

At minimum once per week, the product trio (PM, Designer, Engineer) must:
  1. Talk to a current or potential customer (30-60 minute interview)
  2. Surface new opportunities from what they learned
  3. Map opportunities to the opportunity solution tree
  4. Identify assumptions in proposed solutions
  5. Design small experiments to test those assumptions

This is NOT a research project. It is a weekly ritual.

## Why Continuous Beats Episodic

EPISODIC (typical): 6-week research project → 50-page report → insights already stale
CONTINUOUS: 1 interview/week → insights integrated immediately → decisions based on current data

"Talking to customers once a week is better than a 6-month research project once a year."
— Teresa Torres

Weekly Interview Structure

## The 30-Minute Customer Interview

**Booking**: Recruit 5 customers per week; aim for 1 completed session
**Compensation**: Gift card ($50-100) or product credit

### Opening (5 min)
"Tell me about yourself and how you use [product]."
"What were you trying to accomplish the last time you [core use case]?"

### Story mining (15 min) — PAST ONLY, specific stories
"Walk me through what you did last time you [task]."
"What happened before that? What did you do next?"
"How did that make you feel? What was frustrating about that?"
"What did you try first? Why did that not work?"

Avoid: "What do you usually do?" → hypothetical, not actual
Avoid: "Would you use feature X?" → asking for feature opinions
Seek: Specific stories about what they actually did

### Closing (5 min)
"Is there anything else I should understand about how you work?"
"Who else do you know who faces this challenge?"

### Never ask:
"Do you like this idea?" → confirmation bias
"Would you pay $X for this?" → hypothetical pricing
"What features would you want?" → feature requests ≠ customer problems

Opportunity Solution Tree

## OST Structure

OUTCOME (What business outcome are we trying to achieve?)
  └── OPPORTUNITIES (What customer needs, desires, and pain points stand in the way?)
        ├── Opportunity 1: "Hard to onboard new team members quickly"
        │     └── Solutions: Guided onboarding wizard | Template library | Video walkthroughs
        │           └── Assumptions to test: "Templates will be used within first week"
        ├── Opportunity 2: "Users don't know when tasks are complete"
        │     └── Solutions: Status notifications | Completion animations | Dashboard view
        └── Opportunity 3: "Hard to find recent work"
              └── Solutions: Search | Recent activity feed | Favourites

## OST Rules
- Outcomes come from business strategy (not PMs)
- Opportunities come from customers (not PMs)
- Solutions come from the team (PM + Design + Engineering)
- One outcome per OST (not multiple competing outcomes)
- Opportunities nest (sub-opportunities refine parent opportunities)

Assumption Testing

# Map solutions to their critical assumptions before building
from dataclasses import dataclass
from enum import Enum

class AssumptionType(str, Enum):
    DESIRABILITY  = "desirability"   # Do customers want this?
    VIABILITY     = "viability"      # Will it work for the business?
    FEASIBILITY   = "feasibility"    # Can we build it?
    USABILITY     = "usability"      # Can users use it?

@dataclass
class Assumption:
    solution: str
    assumption: str
    assumption_type: AssumptionType
    importance: int      # 1-10: How important is this assumption?
    confidence: int      # 1-10: How confident are we it's true?

    @property
    def priority_score(self) -> int:
        """High importance + low confidence = test this first."""
        return self.importance * (10 - self.confidence)

# Example: Template library solution
assumptions = [
    Assumption(
        solution="Template library",
        assumption="Users will find pre-built templates relevant to their workflow",
        assumption_type=AssumptionType.DESIRABILITY,
        importance=9, confidence=4,  # High priority to test
    ),
    Assumption(
        solution=,
        assumption=,
        assumption_type=AssumptionType.USABILITY,
        importance=, confidence=,
    ),
]


assumptions.sort(key= a: a.priority_score, reverse=)

Discovery Repository

## Structuring Insights for the Team

After each interview, the PM captures:

**Date**: 2024-03-15
**Customer**: Mid-market SaaS company, Engineering Manager, 50-person team
**Story**: Tried to onboard 3 new engineers last month

**Quote**: "I had to manually recreate the same project setup 3 times — 
there's no way to create a template for it."

**Opportunity**: New team members face repeated setup friction (no templates)

**Assumptions it challenges/confirms**:
- Confirms: Template usage would save significant time (estimated: 2h per hire)
- Challenges: We assumed users wanted to start from scratch; they want to copy

**OST placement**: Opportunity 1.2 (sub-opportunity of "Hard to onboard")

---

Patterns across 10+ interviews → opportunity becomes high confidence
Single interview mention → low signal; track but don't act alone

Anti-Patterns to Avoid

Anti-Pattern	Problem	Fix
Infrequent big research projects	Insights stale; decisions made without customer data	Weekly interview habit; small and continuous
Asking for feature ideas	Customers request solutions, not problems	Ask about past behaviour; surface opportunities
PM interviews alone	Design and engineering miss context	Full product trio in every interview
Insights die in reports	Team reads once; no action	Insights mapped directly to OST; integrated into decisions
Testing too many assumptions at once	Can't attribute which assumption was wrong	One assumption per experiment

10 Rules

1. One customer interview per week minimum — a habit, not a project.
2. The full product trio (PM, Design, Engineering) attends every interview.
3. Ask about past behaviour — never hypothetical future behaviour.
4. Opportunities come from customers; solutions come from the team.
5. The Opportunity Solution Tree is a living document — updated after every interview.
6. Map every proposed solution to its critical assumptions before building.
7. Test the highest-importance, lowest-confidence assumptions first.
8. Experiments are the smallest thing that tests the assumption — not an MVP.
9. Discovery and delivery run in parallel — not sequentially.
10. Insights are shared with the team weekly — discovery is not the PM's private knowledge.

---

Deep Reference Playbook

The sections below extend this skill into a complete operating playbook so it can run end-to-end inside Claude Code, CoWork, or any agentic tool without further prompting. Pull only the sections you need for a given engagement.

Inputs the skill must collect

Before producing any output, the skill confirms:

1. Objective — the single decision or artifact the user wants out of this session.
2. Context — system, team, customer, product, or domain the work sits inside.
3. Constraints — time, budget, headcount, regulatory, technical, political.
4. Definition of done — what "good" looks like and who signs it off.
5. Audience — who reads or consumes the output (engineer, exec, customer, regulator).
6. Existing artifacts — prior versions, related docs, dashboards, tickets.
7. Risk appetite — how reversible the decision is and how much ambiguity is acceptable.

If any of these are missing, the skill asks targeted clarifying questions before generating output. It never invents constraints the user did not state.

Operating workflow

The canonical workflow for Continuous Discovery runs in five stages. Each stage has an explicit exit criterion so the skill knows when to advance.

Stage 1 — Frame. Restate the problem in one paragraph. Name the decision, the deadline, the stakeholders, and the success metric. Surface assumptions explicitly so they can be challenged.

Stage 2 — Diagnose. Inventory the current state with concrete evidence: metrics, quotes, screenshots, configs, tickets. Separate facts from interpretations. Identify the two or three root causes that explain most of the gap, not the long tail of symptoms.

Stage 3 — Design. Generate at least two viable options. For each option, capture: what changes, who owns it, what it costs, what it unblocks, what it risks, and how it could fail. Recommend one with a written rationale.

Stage 4 — Execute. Convert the chosen option into a sequenced plan: milestones, owners, dependencies, gating checks, communication cadence, and rollback triggers. Anything that cannot be assigned an owner and a date is not yet a plan.

Stage 5 — Validate. Define how success will be measured, when the measurement happens, and what action follows each possible result. Schedule the retrospective before the work starts, not after.

Outputs the skill produces

Depending on the request, the skill returns one or more of:

• A one-page brief suitable for an executive reader.
• A detailed working document for the delivery team.
• A decision record capturing the choice, the alternatives, and the rationale.
• A risk register with probability, impact, owner, and mitigation.
• A sequenced action plan with named owners and explicit due dates.
• A measurement plan tied to the success metric.
• A communication plan for stakeholders inside and outside the team.

Every artifact uses clear headings, short paragraphs, and tables where comparison helps. No filler. No restating the prompt. No hedging language when a recommendation is warranted.

Decision logic and trade-offs

The skill applies the following heuristics when choices are not obvious:

• Prefer reversible decisions taken quickly over irreversible decisions taken slowly.
• Optimise for the constraint that bites first — usually time, attention, or trust, not money.
• Default to the simplest design that meets the stated definition of done; add complexity only when a specific requirement forces it.
• Make the cost of being wrong visible so the reader can judge whether the recommendation is proportionate.
• Name the people, not the roles, when assigning ownership; ambiguous ownership produces ambiguous outcomes.

Anti-patterns the skill refuses to emit

Anti-pattern	Why it fails	What the skill does instead
Generic best-practice list with no context	Reader cannot act on it	Tailors recommendations to the stated constraints
Recommendation without trade-offs	Hides the cost of being wrong	Names the price paid for the recommendation
Plan with no owners or dates	Cannot be executed or tracked	Assigns a named owner and a date to every action
Metrics theatre	Measures activity, not outcome	Ties every metric back to the user or business outcome
Boil-the-ocean scope	Nothing ships	Cuts scope to the smallest valuable slice
Buried recommendation	Reader misses the point	Leads with the recommendation in the first paragraph

Quality bar

The skill self-checks each output against these gates before returning it:

1. Can a busy executive understand the recommendation from the first 150 words?
2. Is every claim either evidenced, labelled as an assumption, or removed?
3. Does every action have an owner and a date?
4. Are the trade-offs of the recommendation stated honestly?
5. Is there a measurable success criterion?
6. Would the author be comfortable defending this artifact in a review meeting?

If any gate fails, the skill rewrites the section before returning it.

Worked micro-example

Context: Build a continuous product discovery practice that generates a steady stream of customer insights. Outputs weekly interview cadence, opportunity identification,

Frame: the team needs a defensible recommendation within five working days; the audience is a cross-functional steering group; the cost of delay is higher than the cost of being slightly wrong.

Diagnose: the dominant constraint is decision latency, not analytical depth. Existing data is sufficient for a directional call.

Design: two viable options surfaced. Option A optimises for speed and reversibility. Option B optimises for completeness but slips the deadline by two weeks.

Execute: Option A recommended. Plan sequenced into a two-week sprint with named owners, a mid-point checkpoint, and a clear rollback trigger.

Validate: success measured against a single leading indicator at day 30 and a single lagging indicator at day 90. Retrospective scheduled for day 35.

Cadence and follow-through

A one-shot artifact rarely changes outcomes. The skill recommends a lightweight cadence to keep the work alive:

• Weekly: owner posts a five-line status (done, doing, blocked, risk, ask).
• Fortnightly: steering group reviews leading indicators and unblocks dependencies.
• Monthly: retrospective on what the data is teaching the team; adjust plan accordingly.
• Quarterly: revisit the original objective and decide whether to continue, pivot, or stop.

Closing rules of thumb

1. Lead with the recommendation; supporting analysis follows.
2. Treat every output as a draft that will be challenged; pre-empt the obvious objections.
3. Prefer one strong recommendation over three weak options.
4. When the evidence is thin, say so; do not launder uncertainty as confidence.
5. Optimise for the next decision, not for the perfect document.
6. Make it easy for the reader to disagree with you in a structured way.
7. Ship the artifact; iterate against feedback rather than in private.