Best Practice Guide to Remy Analysis

Overview

Remy has a number of use cases and in this article we'll be exploring best practices for analysis specficially. If you are looking for more of a how-to guide to Remy Analysis, we recommend referencing this article.

Included in this Article

• Before You Start
• The Workflow
  • Orient
  • Investigate
  • Go Deep
  • Build the Story and Package
• Prompt Patterns: Quick Reference
• Tips by Analysis Style
• Quick Start Blueprint

Before You Start

Know your job. The way you approach Remy should reflect what you're trying to accomplish: evaluating concepts, synthesizing insights for a report, building an evidence library, auditing data, or developing recommendations. Your job shapes which prompts matter most and how you use threads.

Use threads intentionally. Each thread maintains its own context. Dedicate a thread to a single concept, topic, or objective. For evidence harvesting, keep a dedicated thread so proof points are consolidated. If a thread gets long and unfocused, start a new one.

1. Orient

Start broad. Before asking about specific themes or quotes, get a map of the terrain.

• What are the top 3 to 5 takeaways from this data about [TOPIC]?
• What, if anything, is surprising or counterintuitive in this data?
• What conversations or data sources are available in this workspace?

Spend time here: ask what segments are represented, how many participants responded, and what questions were asked.

2. Investigate

Turn each research objective into one or two direct prompts so nothing falls through the cracks. Then surface where the data pulls in different directions.

Cover your objectives:

• How do people currently perceive [X]?
• What is working well about [X]? What are the main frustrations?
• What factors seem to drive [OUTCOME] according to participants?

Surface tensions:

• What tensions or trade-offs show up around [TOPIC]?
• On which issues are views most mixed or polarized?

If your project uses a specific framework (customer journey, jobs to be done, NPS drivers), ask Remy to organize findings within it: Map the key themes onto the stages of [FRAMEWORK].

3. Go Deep

Alternate between interpretive questions (what does this mean?) and extraction questions (show me the evidence). Framing without evidence is speculation; evidence without framing is a list.

Explore key themes:

• What drives positive versus negative experiences of [THEME]?
• Approximately how common are concerns related to [THEME]?

Compare segments:

• Are there meaningful differences between [SEGMENT A] and [SEGMENT B] regarding [TOPIC]?
• Which segment appears most at risk, and why?

Extract evidence:

• Pull 5 to 10 representative quotes about [THEME].
• Show contrasting quotes: one very positive, one very negative about [TOPIC].
• Pull quotes from [SEGMENT] about [TOPIC], prioritizing those with high agreement.

Be specific about what you need from quotes: segment, tone, quantity, and purpose. "Give me 5 quotes from managers about onboarding challenges, prioritizing high agreement" will always outperform "give me some quotes."

4. Build the Story and Package

Shape your findings into outputs stakeholders can act on.

• Write 3 to 5 bullets summarizing [TOPIC] for a slide.
• Based on this data, what are your top 3 recommendations for [STAKEHOLDER]?
• For each recommendation, provide the supporting evidence from participant feedback.
• What should we change, what should we preserve, and what risks should we watch?
• Create a comparison table showing how [CONCEPT A] and [CONCEPT B] perform across [DIMENSION 1] and [DIMENSION 2].

Specify the audience and format you need. A slide for leadership looks different from a brief for the research team. If the output isn't quite right, iterate: Make this more concise or Rewrite in a more neutral, findings-focused tone.

Prompt Patterns: Quick Reference

Pattern	What it does	Example
Scoping	Establishes the big picture	What are the top takeaways about [TOPIC]?
Synthesis	Compresses responses into patterns	What themes emerge from feedback on [TOPIC]?
Diagnostic	Explains why something is happening	What drives negative experiences with [X]?
Segmentation	Compares across groups	How does feedback differ between [A] and [B]?
Extraction	Pulls specific evidence	Show me 5 quotes about [THEME] from [SEGMENT].
Evaluation	Judges or prioritizes options	Which concept has the strongest signal, and why?
Packaging	Creates deliverable-ready output	Write 3 bullets summarizing [TOPIC] for a slide.

Productive analysis cycles through these deliberately: scope, synthesize, extract evidence, then package.

Tips by Analysis Style

Concept or Creative Evaluation. Open a separate thread per concept. Follow a summarize-compare-refine loop within each. Use a dedicated comparison thread afterward: Compare the overall reception of [CONCEPT A] and [CONCEPT B].

Insight Synthesis. Work through a broad range of question types. Don't skip the tension prompts; they often produce the most interesting findings. Cycle between framing and evidence throughout.

Evidence Mining. Plan for longer threads. Be specific about selection criteria (segment, agreement level, tone). Ask Remy to include context with each quote: Note which question it came from and which segment the respondent belongs to.

Data Audit. Start with inventory questions: What questions are in this dataset? How many participants responded? When Remy provides a finding, ask about the basis: How many responses is that based on?

Strategy and Recommendations. Complete the orient and investigate stages first. Ask Remy to pair each recommendation with supporting evidence. Pressure-test: What could go wrong if we pursue this?

Quick-Start Blueprint

1. Orient: Top 3 to 5 takeaways and any surprises.
2. Check scope: What conversations, questions, and segments are available.
3. Cover objectives: 1 to 2 prompts per research objective.
4. Surface tensions: Where are views mixed or in conflict?
5. Deep dive: 2 to 4 key themes; explore drivers and prevalence.
6. Segment: Compare across your most important audience groups.
7. Extract evidence: Representative and contrasting quotes for each finding.
8. Shape narratives: Slide-ready bullets or executive summary paragraphs.
9. Define actions: What to change, preserve, and watch.
10. Package: Format for your specific audience and deliverable.