Interview Synthesis

What is interview synthesis?

Interview synthesis is the work you do after customer interviews to turn dozens of pages of notes into a short list of patterns, each one counted and backed by quotes, and then into a decision. Without it, founders remember the 2 most enthusiastic conversations and forget the 8 lukewarm ones, which is how a weak idea survives a whole round of interviews.

The main tool is affinity mapping: you write each observation on its own sticky note and group the notes by similarity, from the bottom up, until themes appear. It is also called the KJ method after Jiro Kawakita, the Japanese anthropologist who developed it in the 1960s to make sense of his field notes. Old and low-tech as it is, it beats re-reading your notes to “get a feel”.

When should you use it?

• After every round of 5 to 15 problem interviews, jobs-to-be-done interviews or shadowing sessions.
• Before you build, write a landing page or pick a segment. Every one of those depends on what the interviews really said.
• When two co-founders remember the same interviews differently. Synthesis settles it with counts.
• For any kind of business: the method doesn’t care whether the notes are about software, a service or a physical product.

Skip it after just 2 or 3 interviews. You won’t have enough to see patterns, and you will start treating single opinions as truths.

How do you do it, step by step?

1. Write your go / pivot / stop criteria BEFORE you look at the notes. Say: “Go if at least 6 of 12 people describe the problem in the last month and at least 3 already pay for a workaround. Pivot if the problem is real but sits with a different group or a different step. Stop if fewer than 4 describe it unprompted.” Writing this first protects you from moving the goalposts once you see the data.

2. Turn every interview into atomic notes. One fact, quote or observation per sticky, with the interviewee code: “Chases vendor invoices on WhatsApp every Friday (I4).” Expect 15 to 25 notes per interview. Facts and quotes only, no conclusions yet. Use one colour per interviewee.

3. Group silently from the bottom up. Put all notes on one board. Pick up a note, find others that feel related and place them together. Don’t create category names first: “pricing”, “features”, “UX” are your boxes, and they hide what customers actually said. If you work with a co-founder, sort in silence for 20 minutes, then talk.

4. Name each cluster with a sentence in the customer’s voice. A label like “Payments” says nothing, while “I lose track of which vendor I’ve paid” points straight at a problem. If you can’t write a clear sentence, the cluster is mixed: split it.

5. Count PEOPLE per cluster. A cluster with 12 notes from one chatty interviewee is 1 person. Write “8 of 14” on each cluster. Then add a mark for intensity: did people show strong emotion, money spent or a workaround? A cluster with high count and high intensity is your real problem.

6. Build a quick matrix in Google Sheets. Rows are interviewees, columns are your top 6 to 8 clusters, cells are 1 or 0. This shows whether some clusters always appear together (one segment) or split cleanly between groups (two segments). That split is a signal for your beachhead segment.

7. Compare against your criteria and write the decision. One paragraph: what you believed, what the counts say, and which of go, pivot or stop you choose. Add the 3 best quotes and the 3 open questions for the next round. Date it.

What does a worked example look like?

Say Hana, a hypothetical founder, believes freelance event planners struggle with clients who pay late. She interviews 14 planners and writes 212 notes.

Her criteria, written first: go if 7 of 14 describe late client payments in the last 2 months and 4 have a workaround.

After 2 hours of grouping, she has 9 clusters. The top ones:

Her original problem fails her own test: 4 of 14, below the line. The vendor side is far stronger. 11 of 14 describe chasing invoices and tracking deposits, and 2 lost real money.

Her decision: pivot within the same customer. She keeps the planners and swaps the problem for tracking what is owed and paid to vendors per event. Next round, she runs 6 contextual inquiry sessions watching planners do their Friday chasing.

Which free tools help?

What mistakes should you avoid?

• Sorting into categories you made up beforehand. Top-down boxes force notes to fit your view. Let the groups emerge from the notes.
• Counting notes instead of people. One talkative interviewee can fill a whole cluster.
• Mixing interpretations into the notes. “Would love our app” is your hope. Write what they said or did.
• Deciding the criteria after seeing the data. The goalposts will move toward your idea. Write them on day 1.
• Waiting until 15 interviews are done. Write your atomic notes within a day of each call, while you still remember tone and context.

How do you know it worked?

You have 3 to 5 named clusters, each with a people count and quotes, a decision written against criteria you set in advance, and a list of open questions for the next round. Our rule of thumb for a go: the top cluster includes at least half your interviewees AND shows high intensity (money spent, workarounds, strong emotion).

A good synthesis also surprises you at least once. If the result matches exactly what you believed before the first interview, check whether you asked leading questions or sorted top-down.

What should you do next?

If the decision is go, write a customer persona and an empathy map from your clusters, then test the promise with a landing page test. A pivot calls for a short new round of interviews on the new problem. And if you’d like an AI co-founder to challenge your clusters and your decision, Fe/male Switch builds that into its startup game for women entrepreneurs.

FAQ

What is affinity mapping in user research?

Affinity mapping is grouping individual notes from interviews or observations by similarity, from the bottom up, until themes appear. Each theme gets a name and a count of how many people it covers. It is also known as the KJ method.

How do you analyse customer interviews quickly?

Write 15 to 25 atomic notes per interview right after each call, then do one 2 to 3 hour sorting session at the end of the round. Count people per cluster and compare against criteria you wrote in advance. That beats re-reading full transcripts.

When should I pivot after customer interviews?

When the problem you tested falls below the bar you set beforehand, but a different problem or a different group shows up strongly in the counts. A pivot keeps what the evidence supports and changes one thing: the problem, the customer or the step in their process.

Updated 26 October 2026 · Written by the Fe/male Switch team

Mark this method as done