How to Synthesize 100 Interviews Without Losing Your Mind

You've done the hard work. You ran 100 interviews. You have hours of recordings, thousands of lines of notes, and a growing sense of dread about making sense of it all.

Synthesis is where most research projects die. This guide will help you survive.

---

Why Synthesis Is So Hard

The Volume Problem

100 interviews = approximately:

• 50-100 hours of recordings
• 500,000+ words of transcripts
• 1,000+ potential insights
• Infinite ways to organize it

No human can hold all this in their head. Your brain isn't designed for it.

The Bias Problem

The interviews you remember best are:

• The most recent ones
• The most emotional ones
• The ones that confirmed what you already believed

This is called availability bias, and it destroys research quality.

The Time Problem

Synthesis takes longer than the interviews themselves. A common ratio:

• 1 hour interview = 2-3 hours synthesis
• 100 interviews = 200-300 hours of synthesis

That's 5-8 weeks of full-time work. Nobody has that.

---

The Progressive Synthesis Method

The key insight: Don't wait until the end to synthesize.

Instead, synthesize progressively as you go. Here's how:

Level 1: During Each Interview

Real-time tagging: As you interview (or immediately after), capture:

• Top 3 pain points mentioned
• Persona indicators
• Surprising quotes
• Key facts (tools used, team size, etc.)

Time: 5-10 minutes per interview

Level 2: After Every 5 Interviews

Pattern check: Look across the last 5 interviews and note:

• Themes appearing multiple times
• Contradictions to investigate
• New questions to add

Time: 30 minutes

Level 3: After Every 20 Interviews

Segment synthesis: Create a mini-report:

• Top 5 pain points so far (with counts)
• Emerging persona segments
• Hypotheses to test in remaining interviews

Time: 2 hours

Level 4: Final Synthesis

By the time you're done, you've already done 80% of the work. Final synthesis is just:

• Consolidating your level 2 and 3 outputs
• Validating patterns hold across all data
• Selecting the best verbatims
• Writing the narrative

Time: 4-8 hours (instead of 200+)

---

The Affinity Mapping Technique

Affinity mapping is the classic synthesis method. Here's how to do it efficiently:

Step 1: Extract Atomic Insights

Go through interviews and extract individual observations onto sticky notes (physical or digital).

Each note should be:

• One idea only
• In the user's voice when possible
• Tagged with source (Interview #, User type)

Example notes:

• "Spends 2 hours/day in meetings" [#12, PM]
• "Tried Notion but team didn't adopt" [#12, PM]
• "Biggest frustration: context switching" [#12, PM]

Step 2: Group by Theme

Move notes into groups of related ideas. Don't pre-define categories—let them emerge.

Common groupings:

• By pain point
• By workflow stage
• By user goal
• By emotion

Step 3: Name the Groups

Once groups are stable, give each a descriptive name that captures the theme.

Good names: "Time lost to unnecessary meetings"

Bad names: "Meetings" (too vague)

Step 4: Count and Prioritize

Count how many notes (and unique users) are in each group. This tells you:

• How widespread is this theme?
• Is it concentrated in one segment or universal?

---

The Synthesis Canvas

Use this template to structure your final output:

| Section | Content | Length | | --- | --- | --- | | Executive Summary | Key finding + recommendation | 3 sentences | | Methodology | Who you talked to, how many, how | 1 paragraph | | Key Findings | Top 5-7 insights with evidence | 1-2 pages | | Persona Summary | Who you found, how they differ | 1 page | | Recommendations | What to do next | 5-10 bullets | | Appendix | All verbatims, full data | As needed |

---

Tools for Synthesis at Scale

Spreadsheet Method

Simplest approach for smaller datasets:

Columns:

• Interview ID
• Date
• User segment
• Pain point 1, 2, 3
• Key quote
• Persona indicators
• Notes

Analysis: Use filters and pivot tables to count patterns.

Tagging Tools

For medium datasets (30-100 interviews):

• Dovetail — Purpose-built for research synthesis
• Notion — Flexible database approach
• Airtable — Spreadsheet + database hybrid
• Miro/FigJam — Visual affinity mapping

AI-Assisted Tools

For large datasets (100+ interviews):

• PulseCheck — Auto-generates synthesis from AI interviews
• Otter.ai — Transcription with AI summaries
• Claude/GPT — Process transcripts with custom prompts

---

The 5-Step Speed Synthesis

When you're short on time, use this compressed method:

Step 1: Skim All Transcripts (2 hours)

Don't read in detail. Skim for highlights. Mark anything that jumps out.

Step 2: Extract Top 3 Per Interview (3 hours)

For each interview, force yourself to pick only the 3 most important observations.

100 interviews × 3 observations = 300 data points (manageable)

Step 3: Cluster into Themes (2 hours)

Group your 300 observations into 10-15 themes. Name each theme.

Step 4: Count and Rank (1 hour)

Count observations per theme. Rank by frequency and importance.

Step 5: Write the Narrative (2 hours)

Turn your top 5-7 themes into a coherent story with evidence.

Total time: 10 hours (vs. 200+ hours for exhaustive synthesis)

---

Avoiding Synthesis Pitfalls

Pitfall 1: Trying to Include Everything

Problem: You end up with 50 "insights" that nobody can act on.

Solution: Force yourself to prioritize. Top 5-7 only. Everything else goes in an appendix.

Pitfall 2: Losing the User's Voice

Problem: Your synthesis is all interpretation, no evidence.

Solution: Every insight needs at least one verbatim quote. Let users speak.

Pitfall 3: Analysis Paralysis

Problem: You keep finding more patterns and never finish.

Solution: Set a deadline. "Good enough" synthesis shipped is better than perfect synthesis never completed.

Pitfall 4: Solo Synthesis

Problem: One person's biases shape everything.

Solution: Involve teammates in clustering or review. Fresh eyes catch blind spots.

---

Key Takeaways

1. Synthesize progressively — Don't wait until the end
2. Use affinity mapping — Let themes emerge from data
3. Force prioritization — Top 5-7 insights only
4. Keep user voice — Verbatims are your evidence
5. Ship imperfect — Done beats perfect
6. Use tools — Scale requires AI assistance

---