Research Product Problems With Six-Layer Analysis

Guides you through a structured, six-layer research process that produces a multi-perspective problem definition before any solution work begins. It examines the problem through customer mental models, ecosystem dynamics, behavioral barriers, past attempts, technology shifts, and feasible scope. Teams that skip this systematic approach confuse symptoms with root causes and build solutions to the wrong problem.

1. You invoke the skill with your product domain and choose Quick (desk research, 5-10 sources per layer) or Full mode (primary research, 25-40+ sources). 2. Claude works through all six layers in sequence, gathering evidence via web search and document analysis, synthesizing findings after each layer. 3. You can pause between layers, ask follow-ups, or re-run specific layers to update individual perspectives. 4. After all six layers complete, the skill produces a unified problem statement, solution direction, and convergence analysis. Output: 14-20 research documents grounded in all six problem-definition lenses.

A problem definition robust enough to survive contact with reality, because you have examined it through six independent lenses before you write a single line of product code.

Requires Claude Code or Claude with code interpreter support because the skill uses WebSearch and WebFetch to gather live research data during the session. The skill is stateless between layers; each layer's output is saved to disk as standalone markdown documents, allowing you to pause mid-framework and resume from where you left off without losing work. Works with any markdown editor and standard file storage.

Quick mode is 2-4 hours per run; Full mode with primary research spans multiple sessions and weeks. Requires you to actively push back on Layer 1 findings if they confirm existing hypotheses (confirmation bias is the most common failure mode). Layer 3 (behavioral dynamics) is the most complex and often reads as surface-level without explicit application of the three frameworks; ask Claude to go deeper if findings feel shallow.