Diamond Develop — Diverge to Create Options

Generate and explore solution options that address the problem statement from Define. This is the first phase of Diamond 2 — the goal is to create a rich option space before converging on the best path in Deliver.

Core Concept

Develop is the creative engine of the engagement. With a clear problem statement and HMW questions from Define, this phase generates multiple possible solutions — not just the obvious one. Good consulting surfaces options the client hadn't considered, challenges "we've always done it this way" thinking, and creates genuine strategic choices.

The key principle: generate before evaluating. Evaluating during generation kills options prematurely — an idea that sounds weak in isolation may become the strongest when combined with another. Develop creates the option space; Deliver evaluates it.

Prerequisites

• Define phase should be complete (problem statement and HMW questions exist)
• Read define/problem-statement.md and define/hmw-questions.md as the brief for this phase

Workflow

1. Load Context

Read consulting-project.json, define/problem-statement.md, and define/hmw-questions.md.

Update phase state:

bash $CLAUDE_PLUGIN_ROOT/scripts/update-phase.sh "<project-dir>" develop in-progress

2. Propose Develop Methods

Read $CLAUDE_PLUGIN_ROOT/references/vision-classes.md for the vision class's recommended Develop methods.

Plugin-powered methods:

Method	Plugin	What It Produces
Value modeling	cogni-trends	TIPS paths from trends to solutions, ranked by business relevance
Proposition modeling	cogni-portfolio	IS/DOES/MEANS messaging for Feature × Market pairs
Solution design	cogni-portfolio	Implementation phases and pricing tiers

Guided methods:

Method	What It Produces	Reference
Scenario planning	2-4 future scenarios with implications	references/methods/scenario-planning.md
Opportunity scoring	Scored option matrix with criteria	references/methods/opportunity-scoring.md
Lean canvas authoring	Research-backed business model hypothesis	references/methods/lean-canvas-authoring.md
Lean canvas refinement	Iterative canvas improvement	references/methods/lean-canvas-refinement.md

Note: Lean canvas methods are recommended for business-model-hypothesis, market-entry, innovation-portfolio, and gtm-roadmap vision classes. For business-model-hypothesis, the lean canvas IS the primary Develop output (replacing proposition modeling and option synthesis). Read $CLAUDE_PLUGIN_ROOT/references/methods/lean-canvas-authoring.md for the full guided workflow.

Ask: "Which methods do you want for option generation? I recommend [2-3 based on vision class]. You can adjust."

3. Value Modeling (cogni-trends)

If the engagement has a tips project from Discovery (check plugin_refs.tips_project):

1. Dispatch cogni-trends:value-modeler on the existing trend candidates
2. The value modeler translates trend candidates into TIPS paths (Trend → Implication → Possibility → Solution)
3. Solutions are ranked by business relevance score
4. Store value model outputs in develop/options/tips-solutions.md

If no tips project exists, offer to run trend-scout first or skip this method.

4. Proposition Modeling (cogni-portfolio)

If the engagement has a portfolio project (check plugin_refs.portfolio_project):

1. Ensure features and markets are defined (from Discovery competitive baseline)
2. Dispatch cogni-portfolio:propositions for Feature × Market pairs
3. Each proposition generates IS (what it is), DOES (what advantage it creates), MEANS (what benefit the buyer gets)
4. Optionally dispatch cogni-portfolio:solutions for implementation phasing
5. Store proposition summaries in develop/propositions/

If no portfolio exists, offer to set one up or skip.

4b. Proposition Quality Gate

If step 4 produced propositions (i.e., develop/propositions/ has content), run a mandatory quality review before continuing. This gate catches vague messaging, unsupported market claims, and broken IS-DOES-MEANS chains before they contaminate Option Synthesis. If no propositions were generated (portfolio was skipped), skip this step.

4b-i. Launch Parallel Persona Review

Launch 2 Task agents in parallel (same turn). Each agent reads:

• develop/propositions/ directory
• define/problem-statement.md (for buyer pain point traceability)
• Discovery competitive baseline (e.g., discover/competitive/summary.md)
• consulting-project.json
• Their persona profile from $CLAUDE_PLUGIN_ROOT/skills/consulting-develop/references/personas/

Persona	Profile	Primary Focus
Proposition Analyst	personas/proposition-analyst.md	IS/DOES/MEANS messaging quality, differentiation, buyer relevance
Market Validator	personas/market-validator.md	Segment precision, competitive position, evidence grounding

Task prompt template:

You are a {PERSONA_NAME} reviewing the Feature x Market propositions of a Double Diamond consulting engagement. Read your persona profile at {PERSONA_PATH} for your evaluation criteria, mindset, and tone.

Read these artifacts:

• Propositions: {PROJECT_DIR}/develop/propositions/ (all files)
• Problem statement: {PROJECT_DIR}/define/problem-statement.md
• Competitive baseline: {PROJECT_DIR}/discover/competitive/summary.md (if exists)
• Engagement config: {PROJECT_DIR}/consulting-project.json

Evaluate each proposition against your 5 criteria. For each criterion, assign PASS/WARN/FAIL with specific evidence. When evaluating competitive distinctness or market positioning, you MUST cross-reference the competitive baseline and cite specific competitors by name. Calculate your weighted score per proposition. Then generate 3-5 questions and identify your most critical concern.

Respond in {LANGUAGE} (technical terms in English).

Agent configuration: Use a fast model (haiku or sonnet), Read tool only.

4b-ii. Evaluate and Decide

After both persona agents complete, read $CLAUDE_PLUGIN_ROOT/skills/consulting-develop/references/proposition-review-protocol.md and apply:

1. Calculate per-persona scores for each proposition
2. Assign per-proposition status:
   • APPROVED: All criteria PASS or WARN — enters Option Synthesis
   • CONDITIONAL: Minor issues noted — enters Option Synthesis with improvements logged
   • BLOCKED: Any FAIL on a criterion weighted >= 25% — excluded from Option Synthesis
3. Identify cross-proposition themes using semantic matching
4. Resolve conflicts: Proposition Analyst wins on messaging quality, Market Validator wins on market fit

4b-iii. Iterate (if needed)

If any propositions are BLOCKED:

1. Identify specific revisions needed (e.g., sharpen IS statement, add competitive evidence to DOES, narrow market segment)
2. Apply revisions to the affected propositions in develop/propositions/
3. Re-run only the persona(s) that flagged FAIL — don't repeat the full review
4. Maximum 2 iteration rounds

After round 2, any still-BLOCKED propositions are excluded from Option Synthesis. Present the exclusion list to the consultant:

Proposition quality gate results:

• N propositions APPROVED
• N propositions CONDITIONAL (improvements noted)
• N propositions BLOCKED and excluded

Blocked propositions: [list with one-line reason each]

You can reinstate any blocked proposition with an explicit rationale. Otherwise, Option Synthesis will proceed with the approved and conditional propositions only.

If the consultant reinstates a blocked proposition, log the override with their rationale in the decision log.

Save results to develop/proposition-review.md.

5. Scenario Planning (Guided)

Read $CLAUDE_PLUGIN_ROOT/references/methods/scenario-planning.md and guide the consultant:

1. Identify 2 critical uncertainties from the problem statement
2. Create a 2×2 matrix using these uncertainties as axes
3. Name and describe 4 resulting scenarios
4. For each scenario, assess: implications for the client, required capabilities, risk profile
5. Map existing options from value modeling against scenarios

Save to develop/scenarios/scenario-matrix.md.

6. Option Synthesis

After all methods complete, synthesize the options. Use only APPROVED and CONDITIONAL propositions from the quality gate (step 4b) — BLOCKED propositions are excluded unless the consultant reinstated them.

1. Consolidate solutions from TIPS value modeling, approved portfolio propositions, and scenario analysis
2. Group into 3-7 distinct strategic options
3. For each option, capture:
   • Name: Short, descriptive label
   • Description: What this option entails
   • Source: Which method surfaced it (TIPS, portfolio, scenario)
   • Alignment: Which HMW question(s) it addresses
   • Key assumptions: What must be true for this to work
   • Quality gate notes (if sourced from a CONDITIONAL proposition): List the outstanding improvements flagged in step 4b that have not yet been addressed. These carry forward as known limitations until Deliver resolves them.
4. Present the option space to the consultant for review

Save to develop/options/option-synthesis.md.

Example option entry (digital-transformation engagement for field service):

Option 3: "Mobile-First Field Platform" Build a unified mobile app replacing 4 legacy field tools. Technicians get real-time job scheduling, parts inventory, and customer history in one interface. Source: TIPS value modeling (mobile workforce trend) + portfolio proposition (field service × mid-market) Alignment: HMW #1 (reduce time-to-resolution) and HMW #3 (improve first-visit fix rate) Key assumptions: Field technicians have reliable mobile connectivity; legacy systems expose APIs for integration.

Present options as equals — ranking happens in Deliver with structured criteria and consultant judgment.

7. Stakeholder Review

Before transitioning to Deliver, stress-test the option space through multi-persona review. This closed-loop quality gate catches issues with feasibility, creative breadth, user grounding, and strategic alignment before the engagement commits to evaluating specific options.

7a. Launch Parallel Persona Review

Launch 4 Task agents in parallel (same turn), one per persona. Each agent reads:

• develop/options/option-synthesis.md
• develop/options/tips-solutions.md (if exists)
• develop/propositions/ directory (if exists)
• develop/scenarios/scenario-matrix.md (if exists)
• define/problem-statement.md and define/hmw-questions.md (for alignment checking)
• consulting-project.json
• The persona's own profile from $CLAUDE_PLUGIN_ROOT/skills/consulting-develop/references/personas/

Personas and their focus:

Persona	Profile	Primary Focus
Engagement Sponsor	personas/engagement-sponsor.md	Strategic alignment, fundability, genuine choice
Solution Architect	personas/solution-architect.md	Technical feasibility, architectural distinctness, constraint adherence
Innovation Strategist	personas/innovation-strategist.md	Creative breadth, assumption challenge, cross-pollination
End-User Advocate	personas/end-user-advocate.md	User value, adoption realism, user-facing risk

Each persona evaluates 5 weighted criteria (weights sum to 100%), assigns PASS/WARN/FAIL verdicts, calculates a weighted score, generates 3-5 stakeholder questions, identifies their most critical improvement, and lists 2-3 key assumptions to test.

Task prompt template:

You are a {PERSONA_NAME} reviewing the Develop phase outputs of a Double Diamond consulting engagement. Read your persona profile at {PERSONA_PATH} for your evaluation criteria, mindset, and tone.

Read these artifacts:

• Option synthesis: {PROJECT_DIR}/develop/options/option-synthesis.md
• TIPS solutions: {PROJECT_DIR}/develop/options/tips-solutions.md (if exists)
• Propositions: {PROJECT_DIR}/develop/propositions/ (if exists)
• Scenario matrix: {PROJECT_DIR}/develop/scenarios/scenario-matrix.md (if exists)
• Problem statement: {PROJECT_DIR}/define/problem-statement.md
• HMW questions: {PROJECT_DIR}/define/hmw-questions.md
• Engagement config: {PROJECT_DIR}/consulting-project.json

Evaluate the option space against your 5 criteria. For each criterion, assign PASS/WARN/FAIL with specific evidence. Calculate your weighted score. Then generate 3-5 questions, identify your most critical improvement, and list 2-3 key assumptions.

Respond in {LANGUAGE} (technical terms in English).

7b. Synthesize & Decide

After all 4 persona agents complete, read $CLAUDE_PLUGIN_ROOT/skills/consulting-develop/references/review-protocol.md and apply:

1. Calculate per-persona weighted scores
2. Identify cross-cutting themes using semantic matching rules
3. Apply priority escalation rules (3+ personas = CRITICAL; specific persona pairs = CRITICAL)
4. Route themes to specific Develop artifacts
5. Resolve conflicts using the tiebreaker hierarchy: Sponsor > End-User Advocate > Solution Architect > Innovation Strategist
6. Merge recommendations by artifact

7c. Iterate (if needed)

If CRITICAL themes are found:

1. Identify specific revisions needed for each affected artifact
2. Apply revisions (e.g., add a stretch option, improve source traceability, add user value articulation)
3. Re-run only the persona(s) that flagged CRITICAL issues — don't repeat the full review
4. Maximum 2 iteration rounds — prevents infinite loops
5. After round 2, present any remaining issues to the consultant for decision regardless of severity

Save review results to develop/review-summary.md.

8. Log and Transition

Update method log and decision log.

Present the Develop summary:

Develop phase complete.

• TIPS solutions generated: N (top 3: ...)
• Propositions modeled: N Feature × Market pairs
• Proposition quality gate: N approved, N conditional, N blocked [N reinstated by consultant]
• Scenarios mapped: 4 (2×2 matrix)
• Strategic options synthesized: N
• Review: [PASSED / PASSED with observations / PASSED after N revision rounds]

Ready to move to Deliver? The final phase will evaluate options, verify claims, build the business case, and generate deliverables.

Mark Develop complete:

bash $CLAUDE_PLUGIN_ROOT/scripts/update-phase.sh "<project-dir>" develop complete

Method Adaptation

For vision-class-specific method recommendations, read $CLAUDE_PLUGIN_ROOT/references/vision-classes.md.

When Things Go Thin

• Only 1-2 options emerge: The divergence was too narrow. Before accepting a thin option space, try these prompts with the consultant: "What would a competitor do?", "What if budget were unlimited?", "What's the opposite of our first option?", "What would we recommend if the constraint on [X] didn't exist?" These reframing questions often unlock options that were implicitly excluded.
• Plugin returns no usable solutions: If value modeling or proposition modeling produces generic or irrelevant output, the input framing may need adjustment. Re-read the problem statement — is it specific enough to generate differentiated solutions? Sometimes the best response is to refine the HMW questions rather than forcing the plugin.
• Consultant fixates on one option early: This is natural but undermines the divergent purpose of this phase. Acknowledge the preferred option explicitly, then say: "Let's develop 2-3 more alternatives so we can compare properly in Deliver. Even if this option wins, the comparison strengthens the recommendation."

Important Notes

• Record why certain options were generated (the reasoning, not just the option) — this traceability matters in Deliver when building the business case
• Cross-reference: if a TIPS solution and a portfolio proposition point to the same thing, note the convergence — it's a signal of robustness
• Scenario planning is particularly valuable for high-uncertainty vision classes where the future state is contested
• If the consultant wants to revisit the problem statement based on what options emerge, that's healthy — the diamond process supports iteration within phases