Diamond Discover — Diverge to Understand

Build a rich, multi-perspective understanding of the problem landscape. This is the first phase of Diamond 1 — the goal is to cast a wide net before converging on a problem statement in the Define phase.

Core Concept

Discover is about breadth, not depth. The consultant and client often arrive with assumptions about what the problem is. This phase deliberately widens the lens — through desk research, trend analysis, competitive mapping, stakeholder input, and data audits — to surface insights that challenge or enrich those initial assumptions.

The key principle: diverge before converging. Premature closure is the enemy of good consulting. Discover builds the evidence base that Define will synthesize.

Prerequisites

• An active diamond engagement (consulting-project.json exists). If not, suggest consulting-setup.
• Read the engagement's vision class and scope from consulting-project.json — they determine which methods are most relevant.

Workflow

1. Load Engagement Context

Read consulting-project.json. Extract: engagement name, vision class, desired outcome, scope, constraints, industry, language.

Update phase state to in-progress:

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

2. Propose Discovery Methods

Read $CLAUDE_PLUGIN_ROOT/references/vision-classes.md for the vision class's recommended Discover methods. Also read any method files referenced.

Present the proposed discovery plan, typically 3-5 activities:

Plugin-powered methods (automated via insight-wave ecosystem):

Method	Plugin	What It Produces
Desk research	cogni-research	Research report with cited sources
Industry trend scan	cogni-trends	60 trend candidates across 4 dimensions × 3 horizons
Competitive baseline	cogni-portfolio	Competitor landscape and market segmentation

Guided methods (interactive prompts with consultant):

Method	What It Produces
Stakeholder mapping	Influence/interest matrix, interview agenda
Data audit	Available data inventory, quality assessment, gaps
Customer journey analysis	As-is journey map with pain points

Method mix check: Before presenting, verify the proposed plan includes at least one internal-facing method (stakeholder mapping, data audit) alongside any external-facing methods (desk research, competitive, trends, customer journey). Discovery that looks only outward misses internal capability constraints, stakeholder dynamics, and data readiness — all of which shape what the Define phase can actually work with. If the vision class recommendations are all external, add stakeholder mapping as a default internal complement.

Ask: "Which methods do you want to use for Discovery? I recommend all plugin-powered methods plus [1-2 guided methods based on vision class]. You can add, remove, or reorder."

3. Execute Plugin Methods

For each confirmed plugin method, dispatch to the appropriate plugin:

Desk Research (cogni-research):

• Frame the research topic from the engagement's desired outcome and scope
• Suggest report type: detailed for most vision classes, deep for digital-transformation or innovation-portfolio
• Recommend market setting matching the engagement scope
• After research completes, store the project path in plugin_refs.research_project
• Copy or symlink the research output summary to discover/research/

Industry Trend Scan (cogni-trends):

• Frame the industry from the engagement context
• Dispatch trend-scout with the industry and language settings
• After scouting completes, store the project path in plugin_refs.tips_project
• Copy or symlink the trend summary to discover/trends/

Competitive Baseline (cogni-portfolio):

• If a portfolio project doesn't exist yet, run portfolio-setup with the client context
• Then dispatch portfolio-scan or compete depending on scope
• Store the project path in plugin_refs.portfolio_project
• Copy or symlink competitive summary to discover/competitive/

Between each plugin dispatch, check with the consultant: "Research complete. Review before moving to trend analysis?"

4. Execute Guided Methods

For each confirmed guided method, read the method file from $CLAUDE_PLUGIN_ROOT/references/methods/ and walk the consultant through it interactively.

Stakeholder Mapping (references/methods/stakeholder-mapping.md):

• Guide the consultant through identifying stakeholders
• Build an influence/interest matrix together
• Draft interview questions aligned to the engagement vision
• Save outputs to discover/stakeholder-map.md

Data Audit (references/methods/data-audit.md):

• Inventory available data sources with the consultant
• Assess quality, recency, and relevance
• Identify critical gaps
• Save outputs to discover/data-audit.md

5. Synthesize Discovery

After all methods complete, produce a discovery synthesis:

1. Read all outputs in discover/ (research summary, trend candidates, competitive data, stakeholder map, data audit)
2. Identify 5-10 key themes that emerge across sources
3. For each theme, cite the specific sources that support it — not just the method name, but the file and section (e.g., *Sources: research/summary.md §Market Size, competitive/summary.md §Market Gaps*). This traceability matters because the Sponsor needs to verify claims for board briefings, and the Define-phase analyst needs to follow evidence trails back to their origin. Vague attribution ("desk research shows...") erodes trust.
4. Note surprises — findings that challenge initial assumptions
5. Flag tensions — contradictions or trade-offs between sources, stating which sources disagree and why the disagreement matters
6. Extract assumptions — for each theme, surface the key assumptions it relies on. Frame them as testable hypotheses: "Assumption: mid-market buyers prioritize speed over features — to be tested via stakeholder interviews in Define." The Define phase needs an explicit assumptions register, not just a theme list. Without it, themes are treated as facts rather than hypotheses, and the engagement builds on unverified ground.
7. Prioritize themes — rank the themes by evidence strength (how many sources support them, how robust the data is) and engagement relevance (how directly the theme connects to the desired outcome). Present the top 3 as "high-confidence, high-impact" themes the engagement should bet on. This helps the consultant and sponsor focus energy where it matters most.
8. Write discover/synthesis.md with sections: Key Themes (with source citations and assumptions), Surprises, Tensions, Assumptions to Test, and Phase Transition Assessment

Example synthesis structure:

### Theme 1: [Specific insight, not a category]
[Evidence-based description integrating multiple sources]
*Sources: research/summary.md §3, competitive/summary.md §Market Gaps, stakeholder-map.md §High Influence*
*Assumption: [testable hypothesis this theme relies on]*
*Priority: HIGH — supported by 3 sources, directly addresses desired outcome*

Present the synthesis to the consultant for review and refinement.

6. Log Methods and Transition

Update the method log:

# For each method used, append to .metadata/method-log.json

Update consulting-project.json with any new plugin_refs.

Ask the consultant: "Discovery phase complete. The synthesis surfaces [N] key themes. Ready to converge in the Define phase, or do you want to explore any theme further?"

If ready to converge, mark Discover complete and suggest consulting-define:

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

Method Adaptation

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

When Things Go Thin

• Plugin returns thin results (fewer than 3 substantive findings): Note the gap explicitly in the synthesis and flag it for the consultant. Thin evidence in one area can be compensated by strength in another, but the consultant should know. Offer to retry with adjusted parameters (broader scope, different market framing) or substitute a guided method.
• Consultant disengages from a guided method: Some methods (stakeholder mapping, data audit) need active input. If the consultant gives minimal responses, capture what's available and move on — a partial output is better than a blocked engagement.
• Prior phase was skipped: Setup should be complete, but if the consultant jumped straight to Discovery, gather the minimum context needed (client, vision class, scope) and create consulting-project.json inline.

Important Notes

• The transition to Define is the consultant's call — they may want to explore further or revisit a finding before converging. Ask, don't assume.
• Sequential dispatch lets findings from each source inform the next — research shapes trend framing, trends inform competitive lens. If the consultant is time-pressed, parallel dispatch works but note the trade-off.
• If a plugin method fails or produces thin results, note the gap and continue — the Define phase can work with incomplete information, and the gap itself is a finding worth recording.
• Update consulting-project.json after each significant step
• Communication Language: Use the engagement's language setting for all interactions