microsoft/hve-core

---
description: 'DT Curriculum Module 3: Synthesis — concepts, techniques, checks, and exercises'
applyTo: '**/.copilot-tracking/dt/**/curriculum-03*'
---

# DT Curriculum Module 3: Input Synthesis

Input synthesis is the Problem Space exit point — the bridge between raw research data and actionable direction for the Solution Space. This module teaches learners how to transform scattered observations, interview notes, and field data into coherent themes that frame problems without prescribing solutions.

## Key Concepts

*Multi-source pattern recognition* — Identifying themes that appear across different types of research data (interviews, observations, environmental audits, existing reports). Patterns that emerge from only one source may be artifacts of research method rather than genuine findings. Learners often anchor on the most vivid or recent data point rather than looking across sources for convergent evidence.

*Theme development progression* — Individual data points become actionable themes through a specific evolution: fragment → supporting evidence from other sources → unified theme → actionable direction. Rushing from fragment to direction produces themes that sound reasonable but lack the evidence base to survive scrutiny. Learners commonly force themes too early, grouping loosely related points under a convenient label.

*Context preservation* — Maintaining domain-specific nuances and environmental factors while abstracting to themes. "Workers struggle with information access" loses critical detail compared to "Night-shift operators spend 10-15 minutes locating manual sections while machines sit idle in 85 dB environments with greasy hands." Learners tend to abstract too aggressively, losing the constraints that make themes actionable.

*Solution-ready problem statements* — Framing discovered themes as clear direction for brainstorming without dictating specific approaches. "Operators need immediate access to repair procedures in hands-free, high-noise environments" enables creative solutions; "Operators need a voice-controlled repair guide" prescribes one. Learners frequently embed solutions in their problem statements without realizing it.

## Techniques

*Affinity clustering* groups individual research data points by natural similarity. Work with physical or virtual cards — one observation per card — and group by emerging themes rather than predetermined categories. Good output is 4-7 theme clusters with clear boundaries. The pitfall is creating categories first and sorting data into them, which forces findings into preconceived frameworks.

*Cross-source validation* tests whether a theme appears in interview data, observation data, and existing metrics. Themes supported by all three carry higher confidence than single-source themes. Good output is a confidence-weighted theme list. The pitfall is discarding themes that appear in only one source without investigating whether the other sources simply did not capture that dimension.

*Stakeholder perspective balancing* ensures synthesis does not over-represent the loudest or most accessible voices. Count how many data points come from each stakeholder group and check for missing perspectives. Good output is a coverage map showing which groups informed which themes. The pitfall is letting management perspectives dominate when frontline workers provided different signals.

## Comprehension Checks

1. A team has 40 interview transcripts and grouped them into 3 themes in one hour. What risks does this speed suggest about their synthesis process?
2. Why does synthesis happen before brainstorming rather than during it? What goes wrong when teams try to synthesize and ideate simultaneously?
3. A synthesis produced the theme "Workers want better technology." Explain what is wrong with this framing and rewrite it as a solution-ready problem statement using manufacturing scenario details.
4. Research found that day-shift operators rarely mentioned information access problems while night-shift operators cited it repeatedly. How should synthesis handle this discrepancy?

## Practice Exercises

*Exercise: Theme from fragments* — Given these manufacturing research fragments, develop one unified theme: (a) "Night-shift workers take 10-15 minutes to find the right manual section," (b) "Day-shift workers ask experienced colleagues instead of using manuals," (c) "Maintenance logs show faster resolution times during shifts with senior operators present." Write the theme as a solution-ready problem statement that preserves environmental context.

*Exercise: Bias check* — Review this stakeholder data distribution and identify the gap: 12 data points from plant managers, 8 from day-shift supervisors, 3 from night-shift operators, 0 from temporary workers. What perspective is missing, why does it matter, and what research method from Module 2 would fill the gap?

## Learner Level Adaptations

Beginners should focus on the difference between premature theme forcing and evidence-based pattern recognition, and practice writing solution-ready problem statements.

Intermediate learners benefit from comparing affinity clustering with top-down categorization and understanding how synthesis quality directly affects brainstorming scope in Method 4.

Advanced learners should explore how organizational power dynamics distort synthesis (whose data gets weighted more), analyze cases where contradictory themes are both valid, and critique the boundary between sufficient synthesis and analysis paralysis.

* All DT coaching artifacts are scoped to `.copilot-tracking/dt/{project-slug}/`. Never write DT artifacts directly under `.copilot-tracking/dt/` without a project-slug directory.