Methodology · Version 1.0

How PatternLab works —
and what it does not.

This page documents PatternLab's methodology, scope, limitations, and appropriate use. It is maintained publicly and updated when our approach changes.

Version 1.0Published March 2026Next review June 2026
01 — What PatternLab is

A pre-research intelligence tool

PatternLab is a tool for surfacing probabilistic experience patterns before human research begins. It synthesizes aggregated public user experience signals into structured archetypes, hypotheses, and research questions — so teams can enter interviews, surveys, and usability studies with sharper frames rather than empty assumptions.

It is designed to support the thinking that happens before research, not to replace the research itself. Its outputs are starting points — not findings.

02 — How it works

Experience pattern synthesis

When a user submits a concept, PatternLab sends the concept name, description, target audience, and known assumptions to Claude — Anthropic's large language model. Claude then reasons across publicly known user experience patterns relevant to the concept domain and returns a structured brief.

01
Concept framing
The user describes their concept, intended audience, and existing assumptions. This framing directly shapes the synthesis.
02
Pattern reasoning
Claude reasons across publicly known experience patterns — drawn from aggregated product reviews, research literature, and public user narratives — relevant to the submitted concept.
03
Archetype construction
Experience archetypes are constructed probabilistically. Each archetype represents a cluster of tendencies, not a type of person. Confidence is assigned based on evidence strength.
04
Hypothesis and question generation
Testable hypotheses and structured research questions are derived from the archetypes — each tied to specific friction points, constraints, or decision tendencies.

All synthesis is performed at the time of submission. PatternLab does not store, train on, or learn from user-submitted concepts.

03 — What the outputs mean

How to read a PatternLab brief

Every PatternLab output is explicitly probabilistic. The following explains what each element means and how it should be used.

Experience Archetypes

Archetypes represent clusters of experience tendencies — not demographics, not segments, not personas. They describe how a type of experience tends to unfold, not who has it. An archetype named "The Workflow-Frustrated" describes a pattern of behaviour, not a category of person.

Confidence Levels

High, Medium, and Low confidence reflect the relative strength of supporting patterns. High confidence means the pattern is well-represented in public experience data. Low confidence means the pattern is speculative and requires validation before acting on it. Confidence is not a measure of accuracy — it is a measure of evidence strength.

Research Hypotheses

Hypotheses are testable propositions derived from archetypes. They are structured to be falsifiable — meaning human research should either confirm or disprove them. A hypothesis is not a finding. It is a question with a provisional answer.

Research Questions

Structured questions are organised by type: objections, confusion, risk, adoption, and comparison. They are designed to be taken directly into user interviews or surveys. They represent the gaps in current knowledge surfaced by the archetypes.

Method Recommendations

Recommended research methods are context-specific suggestions based on the archetypes generated. They are not prescriptive. Research teams should apply their own professional judgement about which methods are appropriate for their specific constraints.

04 — What PatternLab is not

Explicit scope boundaries

These boundaries are not disclaimers. They are design decisions. PatternLab is built to support better research, not to replace it.

-PatternLab does not replace user interviews, usability testing, or statistically representative research
-PatternLab does not simulate real people or predict individual behaviour
-PatternLab does not produce statistically validated findings
-PatternLab does not have access to private user data, internal analytics, or proprietary research
-PatternLab outputs should not be used as the sole basis for product, strategy, or policy decisions
-PatternLab does not guarantee the accuracy or completeness of any output
05 — Appropriate use

When to use PatternLab

Use PatternLab when
+Preparing for user interviews or surveys
+Pressure-testing assumptions before research begins
+Framing hypotheses for discovery sprints
+Aligning teams on what to investigate
+Preparing stakeholder briefs before research
+Exploring a new concept domain quickly
Do not use PatternLab to
-Replace planned user research
-Make product decisions without human validation
-Generate marketing claims about users
-Substitute for accessibility or equity research
-Produce findings for regulatory submissions
-Represent the views of any specific community
06 — Methodology versioning

How we maintain this document

This methodology document is versioned and reviewed quarterly. When our approach changes — whether in synthesis method, output structure, or scope — this page is updated and the version number incremented. Previous versions are retained internally.

Version 1.0
March 2026
Initial methodology published. AI synthesis via Claude. Core output structure established.
07 — Questions and feedback

Get in touch

If you have questions about how PatternLab works, concerns about a specific output, or feedback on this methodology, we want to hear from you. Responsible use matters to us — and that includes hearing about cases where the tool falls short.

Contact
methodology@patternlab.io
We aim to respond within 2 business days.