---
# === IDENTITY ===
id: consulting/recipes/signal-taxonomy-workshop/2026
canonical_question: "How do you run a 2-day signal taxonomy workshop with domain experts?"
aliases:
  - "Signal taxonomy design workshop methodology"
  - "Intent signal classification workshop"
  - "How to build a signal scoring taxonomy with domain experts"
entity_type: execution_recipe
domain: consulting > recipes > Signal Taxonomy Workshop
region: global
jurisdiction: global
temporal_scope: 2026-2027

# === VERIFICATION ===
last_verified: 2026-03-29
confidence: 0.85
version: 1.0
first_published: 2026-03-29

# === TEMPORAL VALIDITY ===
temporal_validity:
  status: evolving
  last_breaking_change: "Initial release — taxonomy workshop methodology v1.0"
  next_review: 2026-09-25
  change_sensitivity: high

# === CONSTRAINTS ===
constraints:
  - "Minimum 1 domain expert required — taxonomy without domain expertise produces unusable classifiers"
  - "Signal audit must be completed before workshop — raw source data required for Day 2 validation"
  - "False positive threshold must be set explicitly — default is 30% max, tighter for regulated industries"
  - "Taxonomy must be validated against minimum 50 real-world examples before sign-off"
  - "Compound signals (multi-source) must be tested independently before combining — correlation != causation"

# === SKIP CONDITIONS ===
skip_this_unit_if:
  - condition: "User needs full Signal Stack lifecycle, not just taxonomy"
    use_instead: "consulting/recipes/signal-stack-engagement-playbook/2026"
  - condition: "User needs signal source identification, not classification design"
    use_instead: "consulting/recipes/signal-source-audit/2026"
  - condition: "User already has taxonomy and needs to build the pipeline"
    use_instead: "consulting/recipes/mvp-pipeline-build/2026"

# === AGENT HINTS ===
inputs_needed:
  - key: signal_audit_status
    question: "Has the signal source audit been completed?"
    type: choice
    options: ["yes — audit report available", "partially — some sources identified", "no — not started"]
  - key: domain_expert_access
    question: "Is a domain expert available for 2-day workshop?"
    type: choice
    options: ["yes — internal expert", "yes — external advisor", "no — need to recruit one"]
  - key: industry_regulation
    question: "How regulated is the target industry?"
    type: choice
    options: ["heavily regulated (healthcare, finance)", "moderately regulated (construction, food)", "lightly regulated (SaaS, consulting)"]

# === EXECUTION METADATA ===
execution:
  required_inputs:
    - name: "Signal Source Audit Report"
      source: "consulting/recipes/signal-source-audit"
      format: "spreadsheet + document"
    - name: "Domain expert availability"
      source: "client/strategy"
      format: "calendar confirmation"
    - name: "Sample raw data from top 5 sources"
      source: "data team"
      format: "CSV/JSON"

  outputs:
    - name: "Signal Taxonomy Document"
      format: "JSON schema + document"
      description: "Hierarchical classification of intent signals with weighted scoring model"
    - name: "Scoring Calibration Dataset"
      format: "spreadsheet"
      description: "50+ scored real-world examples used for taxonomy validation"
    - name: "False Positive Analysis"
      format: "document"
      description: "Analysis of false positive rate and threshold recommendations"

  tools_required:
    - name: "Workshop facilitation tools"
      purpose: "Structured brainstorming and decision-making"
      tier: "free"
      cost: "$0"
      alternatives: ["Miro", "FigJam", "physical whiteboard"]
    - name: "Spreadsheet"
      purpose: "Signal scoring and validation tracking"
      tier: "free"
      cost: "$0"
      alternatives: ["Google Sheets", "Excel", "Airtable"]
    - name: "Sample data access"
      purpose: "Live validation against real data"
      tier: "varies"
      cost: "$0-$500 (sample API calls)"
      alternatives: []

  credentials_needed: []

  estimated_duration: "2 days (workshop) + 3 days (validation)"
  estimated_cost: "$3K-$8K (consulting time)"

# === DISTRIBUTION ===
canonical_source: "https://knowledgelib.io/consulting/recipes/signal-taxonomy-workshop/2026"
suggested_citation: "Source: knowledgelib.io — AI Knowledge Library (verified 2026-03-29)"

# === RELATED UNITS ===
related_kos:
  depends_on:
    - id: "consulting/recipes/signal-source-audit/2026"
      label: "Signal audit outputs required as workshop input"
  feeds_into:
    - id: "consulting/recipes/mvp-pipeline-build/2026"
      label: "Taxonomy feeds directly into pipeline classifier"
  related_to:
    - id: "consulting/recipes/signal-stack-engagement-playbook/2026"
      label: "Master playbook — this is Phase 2"

# === SOURCES ===
sources:
  - id: src1
    title: "Taxonomy Design for Information Architecture"
    author: Rosenfeld, L. & Morville, P.
    url: https://www.oreilly.com/library/view/information-architecture-4th/9781491911686/
    type: industry_report
    published: 2015-09-25
    reliability: authoritative
  - id: src2
    title: "Thinking, Fast and Slow — Heuristics and Biases"
    author: Kahneman, D.
    url: https://www.penguinrandomhouse.com/books/89308/thinking-fast-and-slow-by-daniel-kahneman/
    type: academic_paper
    published: 2011-10-25
    reliability: authoritative
  - id: src3
    title: "The Design Sprint — Methodology"
    author: Knapp, J., Zeratsky, J. & Kowitz, B.
    url: https://www.thesprintbook.com/
    type: industry_report
    published: 2016-03-08
    reliability: high
  - id: src4
    title: "Signal Detection Theory and ROC Analysis"
    author: Macmillan, N.A. & Creelman, C.D.
    url: https://www.cambridge.org/core/books/detection-theory/3B8C9980B13E2391B3F31715E06F5AB4
    type: academic_paper
    published: 2004-11-01
    reliability: authoritative
  - id: src5
    title: "Facilitator's Guide to Participatory Decision-Making"
    author: Kaner, S.
    url: https://www.wiley.com/en-us/Facilitator%27s+Guide+to+Participatory+Decision+Making-p-9781118404959
    type: industry_report
    published: 2014-04-14
    reliability: high
---

# Signal Taxonomy Workshop

## Purpose

This recipe executes a structured 2-day workshop that transforms raw signal source data into a validated classification taxonomy. Day 1 focuses on knowledge extraction from domain experts — mapping trigger events, identifying signal patterns, and establishing business context. Day 2 focuses on quantification — scoring signal strength, setting false positive thresholds, designing compound signals, and validating the taxonomy against live data. [src1, src3]

## Prerequisites
<!-- Agents: verify ALL prerequisites before executing. Missing prerequisites = failed execution. -->

- [ ] **Signal Source Audit** completed — scored inventory of available data sources
- [ ] **Domain expert** confirmed for 2 full days — internal industry specialist or external advisor
- [ ] **Sample raw data** pulled from top 5 sources — CSV/JSON format for live validation
- [ ] **Workshop space** booked — whiteboard, projector, internet access for live data pulls
- [ ] **Facilitator** assigned who understands both signal methodology and workshop facilitation

## Constraints
<!-- Hard rules. Agents: enforce throughout execution. Violating these = broken output or legal risk. -->

- At least 1 domain expert must participate — taxonomy designed without domain expertise produces classifiers that miss industry-specific nuances. [src2]
- Signal source audit must be completed before the workshop. Day 2 requires real data for validation.
- False positive threshold must be explicitly set during the workshop. Default maximum: 30%. For regulated industries (healthcare, finance): 15-20%. [src4]
- Taxonomy must be validated against minimum 50 real-world examples before final sign-off.
- Compound signals must be tested independently before combining — observed correlation between signals does not imply they compound reliably.

## Tool Selection Decision

```
Which workshop format?
├── In-person (recommended)
│   └── PATH A: Full 2-day workshop — whiteboard, live data, real-time iteration
├── Remote synchronous
│   └── PATH B: 2-day virtual workshop — Miro/FigJam, screen sharing, structured breaks
├── Hybrid (domain expert remote)
│   └── PATH C: Facilitator on-site, expert via video — requires strong facilitation
└── Async (domain expert unavailable for 2 days)
    └── PATH D: 4x half-day sessions over 2 weeks — slower but accommodates schedules
```

| Path | Format | Cost | Speed | Output Quality |
|------|--------|------|-------|---------------|
| A: In-person | Full 2-day on-site | $5K-$8K | 5 days total | Excellent |
| B: Remote | 2-day virtual | $3K-$5K | 5 days total | Good |
| C: Hybrid | Mixed on-site/remote | $4K-$6K | 5 days total | Good |
| D: Async | 4 half-day sessions | $3K-$5K | 10-14 days | Adequate |

## Execution Flow

### Step 1: Pre-Workshop Preparation (Day 0)

**Duration**: 1 day
**Tool**: Document preparation + data staging

Prepare workshop materials:
- Print signal source audit report (all participants)
- Load sample data from top 5 sources into shared spreadsheet
- Create taxonomy template with blank hierarchy (signal category > signal type > specific indicator)
- Prepare 20+ real company examples from target vertical for validation exercise
- Draft agenda and circulate to all participants [src3]

**Verify**: All materials prepared, sample data accessible, participants confirmed.
**If failed**: If sample data unavailable, workshop can proceed with audit report only, but Day 2 validation will be weaker.

### Step 2: Day 1 Morning — Domain Expert Interviews (3 hours)

**Duration**: 3 hours
**Tool**: Structured interview + whiteboard mapping

Conduct semi-structured interviews with domain expert(s):
- What events in this industry indicate a company is about to need [target service/product]?
- What behavioral changes would you observe 3-6 months before a purchase decision?
- What public data would tell you a company is struggling with [target problem]?
- What are the false signals — things that look like intent but are not?

Map responses onto whiteboard as a preliminary signal hierarchy. Group related signals into categories (regulatory, behavioral, financial, operational). [src1, src5]

**Verify**: Minimum 20 distinct signal types identified across at least 3 categories.
**If failed**: If fewer than 15 signals, domain expert may need prompting with specific source examples from the audit report.

### Step 3: Day 1 Afternoon — Data Source Deep-Dive (3 hours)

**Duration**: 3 hours
**Tool**: Signal audit report + source-by-source walkthrough

Walk through each data source from the audit report with the domain expert:
- For each source: What specific signals can we extract? How strong are they individually?
- Map each source to the signal hierarchy from Step 2
- Identify gaps: signals we want but have no source for
- Identify compound opportunities: which signals become stronger when combined?
- Document trigger event sequences — what happens first, second, third before a purchase

**Verify**: Signal hierarchy mapped to data sources. Compound signal opportunities documented. Gap analysis complete.
**If failed**: If domain expert unfamiliar with specific data sources, demonstrate with sample data and ask for interpretation.

### Step 4: Day 2 Morning — Signal Strength Scoring (3 hours)

**Duration**: 3 hours
**Tool**: Scoring framework + spreadsheet

Assign strength scores to each signal type:
- **Individual signal strength** (1-10): How predictive is this signal alone?
- **Reliability** (1-5): How consistently available and accurate is the underlying data?
- **Timeliness** (1-5): How far in advance of purchase does this signal appear?
- **Compound multiplier** (1.0-2.0): How much stronger when combined with correlated signals?

Set false positive threshold: what percentage of flagged companies turn out to be non-buyers? Default max: 30%. Regulated industries: 15-20%. [src2, src4]

**Verify**: All signals scored. False positive threshold explicitly set and documented.
**If failed**: If consensus on scoring cannot be reached, use median scores and document dissenting views for post-workshop calibration.

### Step 5: Day 2 Afternoon — Live Validation (3 hours)

**Duration**: 3 hours
**Tool**: Sample data + taxonomy + scoring model

Validate taxonomy against real data:
- Pull 50+ company records from sample data
- Apply taxonomy scoring to each company
- Compare automated classification against domain expert judgment
- Calculate false positive rate and false negative rate
- Identify edge cases and refine classification rules
- Document exceptions and special handling rules [src4]

**Verify**: Taxonomy validated against 50+ examples. False positive rate within threshold. Edge cases documented.
**If failed**: If false positive rate exceeds threshold, tighten scoring weights, add qualifying signals, or narrow signal definitions. Revalidate with fresh sample.

### Step 6: Post-Workshop Documentation (Days 3-5)

**Duration**: 2-3 days
**Tool**: Document generation + JSON schema creation

Produce final taxonomy deliverables:
- Signal Taxonomy Document: hierarchical classification with all scoring weights
- JSON schema for pipeline implementation (machine-readable taxonomy)
- Scoring Calibration Dataset: all 50+ validation examples with scores and outcomes
- False Positive Analysis: rate, contributing factors, recommended monitoring approach
- Implementation notes: edge cases, special rules, domain expert caveats

**Verify**: All deliverables reviewed by domain expert and engagement lead. JSON schema parseable and complete.
**If failed**: If domain expert unavailable for review, flag taxonomy as "provisional" and schedule 1-hour review call within 5 business days.

## Output Schema

```json
{
  "output_type": "signal_taxonomy",
  "format": "JSON schema + document",
  "sections": [
    {"name": "signal_hierarchy", "type": "object", "description": "Category > type > indicator tree with scoring weights", "required": true},
    {"name": "scoring_model", "type": "object", "description": "Strength, reliability, timeliness, compound multiplier per signal", "required": true},
    {"name": "false_positive_threshold", "type": "number", "description": "Maximum acceptable false positive rate", "required": true},
    {"name": "compound_signals", "type": "array", "description": "Multi-source signal combinations with multipliers", "required": true},
    {"name": "validation_dataset", "type": "array", "description": "50+ scored examples with outcomes", "required": true},
    {"name": "edge_cases", "type": "array", "description": "Exceptions and special handling rules", "required": true}
  ],
  "expected_sections": "6",
  "sort_order": "signal strength descending"
}
```

## Quality Benchmarks

| Quality Metric | Minimum Acceptable | Good | Excellent |
|---------------|-------------------|------|-----------|
| Signal types identified | > 20 | > 30 | > 50 |
| Validation examples tested | 50 | 75 | 100+ |
| False positive rate | < 30% | < 20% | < 10% |
| Signal categories covered | 3 | 4 | 5+ |
| Compound signals designed | 2 | 5 | 10+ |

**If below minimum**: Extend workshop by half day or schedule follow-up session with additional domain expert input.

## Error Handling

| Error | Likely Cause | Recovery Action |
|-------|-------------|----------------|
| Domain expert unavailable | Schedule conflict | Reschedule or switch to async format (Path D) |
| Fewer than 15 signals identified | Narrow vertical or inexperienced expert | Prompt with audit report examples; consider second expert |
| False positive rate > 40% | Taxonomy too broad or signals too noisy | Tighten definitions, add qualifying criteria, revalidate |
| No compound signals found | Signals are independent | This is valid — document and proceed with individual signals |
| Expert and facilitator disagree on scoring | Different assumptions about buyer behavior | Document both perspectives, test empirically during pilot |

## Cost Breakdown

| Component | Remote ($3K-$5K) | In-Person ($5K-$8K) | Deep ($7K-$12K) |
|-----------|-------------------|---------------------|-----------------|
| Pre-workshop prep | $500-$800 | $800-$1.2K | $1.2K-$2K |
| Day 1: Domain interviews | $800-$1.2K | $1.2K-$2K | $2K-$3K |
| Day 2: Scoring + validation | $800-$1.2K | $1.2K-$2K | $2K-$3K |
| Post-workshop documentation | $500-$800 | $800-$1.2K | $1.2K-$2K |
| Domain expert compensation | $400-$800 | $800-$1.5K | $1.5K-$2K |
| **Total** | **$3K-$5K** | **$5K-$8K** | **$7K-$12K** |

## Anti-Patterns

### Wrong: Designing taxonomy without domain expertise
Building signal classification from data patterns alone without industry expert validation. Result: classifier flags the wrong companies — high false positive rate, sales team loses trust within first week. [src2]

### Correct: Domain expert drives taxonomy design
Minimum 1 domain expert participates for full 2 days. Their industry knowledge catches false signal patterns that data analysis alone misses.

### Wrong: Skipping live data validation
Finalizing taxonomy based on theoretical signal design without testing against real data. Result: taxonomy looks elegant on paper but fails on first real dataset. [src4]

### Correct: Validate against 50+ real examples before sign-off
Day 2 afternoon is dedicated to applying the taxonomy to real data. If it fails validation, iterate before leaving the workshop.

### Wrong: Setting implicit false positive thresholds
Not discussing acceptable error rates during the workshop. Result: pipeline delivers too many irrelevant dossiers, overwhelming sales team. [src4]

### Correct: Explicitly set and document the false positive threshold
Discuss and agree on maximum acceptable false positive rate during Day 2. Write it into the taxonomy document as a hard constraint.

## When This Matters

Use when an agent needs to design an intent signal classification system for a specific industry vertical. This is Phase 2 of the Signal Stack engagement — it transforms raw signal source data into a structured, validated taxonomy that can be implemented as an automated classifier. Requires completed signal source audit and domain expert participation.

## Related Units

- [Signal Stack Engagement Playbook](/consulting/recipes/signal-stack-engagement-playbook/2026)
- [Signal Source Audit](/consulting/recipes/signal-source-audit/2026)
- [MVP Pipeline Build](/consulting/recipes/mvp-pipeline-build/2026)