> ## Documentation Index
> Fetch the complete documentation index at: https://docs.hallwayai.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Synthesis Reports

> Understand the components of Hallway's AI-generated research synthesis reports

Synthesis reports combine insights from all your processed interviews into a cohesive analysis. This guide explains each component of the report and how to get the most value from your research data.

## Report Overview

A synthesis report contains five main sections:

<CardGroup cols={2}>
  <Card title="Statistics Bar" icon="chart-simple">
    At-a-glance metrics for interviews, themes, findings, and quotes.
  </Card>

  <Card title="Summary" icon="file-lines">
    Executive overview of key findings across all interviews.
  </Card>

  <Card title="Key Findings" icon="lightbulb">
    Specific discoveries organized by learning objective.
  </Card>

  <Card title="Emerging Themes" icon="quote-left">
    Patterns with supporting quotes and visual distribution chart.
  </Card>

  <Card title="Metadata" icon="circle-info">
    Interview count, version, and last updated timestamp.
  </Card>
</CardGroup>

## Statistics Overview

At the top of every synthesis report, you'll find a statistics bar that provides key metrics at a glance:

| Metric           | Description                                 |
| ---------------- | ------------------------------------------- |
| **Interviews**   | Total number of processed sessions included |
| **Themes**       | Number of emerging themes identified        |
| **Key Findings** | Number of specific insights discovered      |
| **Quotes**       | Total supporting quotes across all themes   |

<Tip>
  The statistics bar uses color-coded cards for quick visual scanning—navy for interviews, purple for themes, teal for findings, and pink for quotes.
</Tip>

## Report Metadata

The metadata card at the top of your synthesis provides context about the data:

| Field          | Description                                             |
| -------------- | ------------------------------------------------------- |
| **Interviews** | Number of processed sessions included in this synthesis |
| **Updated**    | Date and time the synthesis was last generated          |
| **Version**    | Increments each time you regenerate the synthesis       |

<Note>
  Only processed sessions are included. Completed but unprocessed sessions won't appear in the synthesis.
</Note>

## Summary Section

The summary provides a high-level overview suitable for stakeholders who need quick insights without reading the full report.

### What the Summary Includes

* **Key takeaways**: The most important findings in 2-3 paragraphs
* **Pattern highlights**: Major themes that emerged across interviews
* **Actionable insights**: Recommendations based on the research

### Using the Summary Effectively

The summary is formatted in Markdown and renders with proper headings, bullet points, and emphasis. It's designed to:

* Stand alone as an executive briefing
* Provide context before diving into detailed findings
* Be copy-pasted into presentations or documents

<Tip>
  The summary is regenerated each time you update the synthesis. As you add more interviews, the summary evolves to reflect the fuller picture.
</Tip>

## Key Findings

Key findings are specific discoveries tied to your learning objectives. Each finding represents a concrete insight that addresses what you set out to learn.

### Finding Structure

Each key finding contains:

```
Finding
├── Related Objective (if applicable)
├── Finding Statement
└── Evidence
    ├── Evidence point 1
    ├── Evidence point 2
    └── Evidence point 3
```

### Understanding Related Objectives

When a finding maps to one of your learning objectives, you'll see the objective displayed above the finding. This helps you track how well your research addressed each objective.

**Example:**

> *Related Objective: Understand what triggers users to seek a solution like ours*
>
> **Finding:** Users typically start searching for solutions after experiencing repeated frustration with manual workarounds, often triggered by a specific incident that costs significant time or causes an error.

### Evidence Points

Evidence points are supporting details extracted from interview transcripts. They're not direct quotes but rather paraphrased observations that support the finding.

<Tabs>
  <Tab title="Strong Evidence">
    * Specific, concrete observations
    * References particular user behaviors
    * Quantifies when possible

    *Example:* "7 of 10 respondents mentioned trying spreadsheets first before seeking a dedicated solution."
  </Tab>

  <Tab title="Weak Evidence">
    * Vague or generic statements
    * Doesn't tie to specific behaviors
    * Lacks supporting detail

    *Example:* "Users don't like the current process."
  </Tab>
</Tabs>

## Emerging Themes

Themes represent patterns that appeared across multiple interviews. Unlike key findings (which map to specific objectives), themes emerge organically from the data.

### Theme Distribution Chart

A donut chart at the top of the themes section visualizes how quotes are distributed across themes. This helps you quickly see:

* **Relative weight**: Which themes have the most supporting evidence
* **Total quotes**: The center of the chart shows the total quote count
* **Color coding**: Each theme segment matches its card color

<Note>
  The chart is accessible to screen readers with a full description of the data.
</Note>

### Theme Components

Each theme includes:

| Component             | Description                                                          |
| --------------------- | -------------------------------------------------------------------- |
| **Color Accent**      | Visual indicator matching the chart (navy, teal, purple, pink, blue) |
| **Name**              | Short, descriptive label (e.g., "Fear of Data Loss")                 |
| **Description**       | Explanation of what the theme represents                             |
| **Quote Count**       | Number of supporting quotes displayed as a badge                     |
| **Supporting Quotes** | Direct quotes from respondents                                       |

### Color-Coded Theme Cards

Each theme card has a unique color accent on its left border that corresponds to its segment in the distribution chart. This color coding:

* **Improves scannability**: Quickly identify and reference specific themes
* **Creates visual consistency**: The same color appears in the chart and card
* **Enhances quotes**: Expanded quotes have a tinted background matching the theme color

### Exploring Themes

Themes are collapsible. Click on any theme to expand it and see all supporting quotes:

* **Quote text**: The exact words from the respondent in italics
* **Attribution**: Which respondent said it (prefixed with "—")

<Tip>
  Supporting quotes are powerful for stakeholder presentations. They bring research to life with real voices.
</Tip>

### Theme vs. Finding

| Themes                      | Findings                     |
| --------------------------- | ---------------------------- |
| Emerge from the data        | Tied to learning objectives  |
| Pattern-focused             | Insight-focused              |
| Include direct quotes       | Include paraphrased evidence |
| Describe "what's happening" | Answer "what did we learn"   |

Both are valuable—themes show patterns while findings answer your research questions.

## Report Versioning

Each time you regenerate the synthesis, the version number increments:

* **Version 1**: Initial synthesis generation
* **Version 2**: First regeneration (after new sessions or changes)
* **Version 3+**: Subsequent regenerations

<Warning>
  Previous versions are not stored. If you need to preserve a version, export it before regenerating.
</Warning>

## Stale Reports

A synthesis becomes "stale" when the underlying data changes:

### When Reports Go Stale

* Sessions have been deleted since the last synthesis
* This is indicated by an amber warning banner

### What Stale Means

The current synthesis may include insights from sessions that no longer exist. The findings might reference quotes or evidence from deleted data.

### Resolving Stale Reports

Click **Regenerate** to create a fresh synthesis based on current data. The new synthesis will only include insights from sessions that currently exist.

## Report Quality

The quality of your synthesis depends on several factors:

<AccordionGroup>
  <Accordion title="Number of interviews">
    More interviews generally produce richer, more reliable synthesis. With only 1-2 interviews, findings are preliminary. With 10+, patterns are more robust.
  </Accordion>

  <Accordion title="Interview quality">
    Longer, more in-depth interviews produce better insights. Short or surface-level conversations may not provide enough material for meaningful synthesis.
  </Accordion>

  <Accordion title="Learning objective clarity">
    Clear, specific objectives lead to more actionable findings. Vague objectives produce vague findings.
  </Accordion>

  <Accordion title="Respondent diversity">
    Diverse respondents reveal different perspectives. If all respondents are similar, themes may be narrower.
  </Accordion>
</AccordionGroup>

## Using Reports Effectively

### For Your Team

* Share the summary for quick alignment
* Discuss key findings in research reviews
* Use themes to identify areas for deeper exploration

### For Stakeholders

* Lead with the summary for executives
* Highlight 2-3 key findings most relevant to decisions
* Include supporting quotes for credibility

### For Product Decisions

* Map findings to specific features or changes
* Prioritize based on theme frequency and intensity
* Validate findings with additional research when needed

## Regenerating Reports

You should regenerate your synthesis when:

<Tabs>
  <Tab title="New Data">
    * New sessions have been processed
    * You want insights to reflect all available data
    * You've hit a milestone (5, 10, 15 interviews)
  </Tab>

  <Tab title="Data Changes">
    * Sessions have been deleted
    * You see the "stale synthesis" warning
    * The current report references removed data
  </Tab>

  <Tab title="Quality Check">
    * You want to verify findings with fresh analysis
    * You're about to share with stakeholders
    * You're preparing the final research report
  </Tab>
</Tabs>

## Next Steps

* [Exporting Data](/docs/help/insights/exporting-data) - Share reports with stakeholders
* [Understanding Insights](/docs/help/insights/understanding-insights) - How the insights system works
* [Creating Projects](/docs/help/getting-started/creating-projects) - Improve your learning objectives