The Interview Log & Evidence Rules

By Kasper-Tanakrit Sermsuksan · Global Venture I: Playbook · Block C

You are here: C-05 · Block C · Understanding the Market · article 5 of 9 · Form C · Next: C-06 First 10 Customers & Angel Customer™

In short: In a customer interview, evidence is what the person did or committed to (facts, past behavior, commitments), not what they said they might do — and the strongest logs show that evidence changed the idea. Opinions, compliments and "I would definitely buy that" are not evidence. Log every interview the same day, in your own words, tag each note by type, and keep one running list of all your interviews. Ask before photos or recordings, delete recordings once you no longer need them, and protect what people share with you.

Try it in 10 minutes: Think of the last real conversation you had about your idea, or any conversation from today. Write down five things the other person said. Tag each one F (fact), B (past behavior), C (commitment) or O (opinion). If most are O, rewrite one of your questions so it would have produced an F.

Your first interviews will feel great.

People are kind to students who ask nicely. They nod. They say "That's a great idea." Some say they would definitely buy it.

You walk out feeling validated. Then you sit down to write your notes and find almost nothing you can use.

One student from the 2025 class wrote in their end-of-course reflection that they finally understood why their earlier startup ideas never felt valid: they had not been building what customers really wanted. Talking to customers, they wrote, is what makes a startup real.

Not ideas. Not compliments. Evidence.

So here is how I tell evidence from noise. And how I want you to write it down, so it is still useful a month from now.

What counts as evidence from a customer interview?

To me, three things count. Everything else is decoration.

• Facts — what actually happened. "Last Tuesday my father fell in the bathroom. I took the afternoon off and we waited four hours at the hospital."
• Past behavior — what they already did, tried, spent or paid. "I've tried three exercise apps for him. I paid for one. We stopped after two weeks."
• Commitments — what they gave up for you: their time (a second meeting, a trial), their reputation (an introduction to their boss or friends), or their money (a deposit, a pre-order, a paid pilot). Rob Fitzpatrick, in The Mom Test, calls these the real currency of early conversations: "The major currencies are time, reputation risk, and cash"¹.

What does not count as evidence?

• Compliments — "That's a great idea!" Pleasant. Useless. People are kind to strangers who ask nicely.
• Opinions — "I think most people would want that." About other people, about the future, about things they have not done.
• Hypotheticals — "I would definitely buy that." "I'd pay maybe ฿500 a month." Promises about the future are the least reliable thing anyone will tell you.
• Your own interpretation written as if they said it. "She clearly needs our app." No. She said she worries during the workday. Write that.

Of all of these, "I would definitely buy that" is the line I trust least.

I use a simple test for every note: did this happen, did they do it, or did they give something up? If none of the three, tag it as opinion — and do not let it carry weight in your decisions.

Opinions are not worthless. They can point you toward what to ask next.

They just cannot prove anything.

How do you log a customer interview?

Log it the same day. Before you check your phone. Before you talk to your team. Before your memory quietly edits what you heard.

I recommend the same structure every time, so all your interviews can be compared side by side.

What should a customer interview log template include?

These are the fields I use, every time.

• Date · type: When, and what kind of interview — for example entrepreneur, customer, ASEAN friend, cross-discipline friend
• Who: Role and context, not full name unless they agreed ("working daughter, 42, Bang Na; father 78")
• How you reached them: Channel and who introduced you — this becomes your outreach map. For customers and entrepreneurs, I recommend strangers; an introduction from family is fine.
• Consent: What they agreed to: notes only / photo / recording / quoting by name
• Facts: What happened, with dates and numbers where they gave them
• Past behavior: What they tried, used, spent or paid
• Exact words: One to three phrases, word for word, in quotation marks — the language you will reuse
• Commitments: Time, reputation or money they offered — or "none"
• Opinions: Kept separate, clearly tagged
• Biggest surprise: The one thing you did not expect
• What changed: What this interview changes in your idea, problem statement or beachhead — or "nothing yet"
• Proof: A photo together, with permission — or your thank-you email if they declined a photo
• Next: Follow-up, introductions offered, thank-you sent within 24 hours (date)
• AI used for: What, if anything, AI helped with — never the notes themselves

Then add one line to a running list of all your interviews, with a link to the entry. Over time, those lines become the backbone of whatever you write next: a report, a deck, a memo.

Write it in your own words

I recommend writing your notes yourself. AI is welcome for research and drafting. But there is a practical reason to keep it out of your notes.

The moment you paste notes into a tool and ask it to "summarize the insights," the surprise disappears. AI is very good at making things sound reasonable.

Insight usually sounds unreasonable at first.

And never paste someone's personal details into an AI tool without their permission.

How do you judge the quality of interview evidence?

When I read interview logs, I use three bars:

• A — What it means: Real evidence from real people changed the idea, and it shows — What it looks like: "After customers 2 and 4 both described paying for physio but quitting home exercises, we moved our beachhead from older people living alone to their working children."
• B — What it means: Complete, with some evidence — What it looks like: All fields filled; some facts and past behavior; the link to a decision is weak
• C — What it means: Complete, mostly opinion — What it looks like: "Everyone loved the idea and said they would use it."
• Not yet — What it means: Missing parts — What it looks like: No log, no consent note, or interviews not yet done

Notice that A is not about how many people liked your idea. It is about whether evidence changed something.

Sometimes the strongest entry I read is the one that says, "This killed our first idea."

What are the consent and privacy rules for interviews?

People who talk to you are trusting you with their stories, sometimes their health, their money worries or their family. I ask you to treat that seriously.

• Ask before you record or photograph. Every time. "Is it okay if I take notes on my laptop?" is a good start; recordings and photos need a clear yes.
• Proof of the interview: a photo, with permission. If someone declines a photo, respect that — your thank-you email can be the record.
• Do not publish names without permission. In your report and deck, use roles ("a pharmacy owner in Khon Kaen") unless they agreed to be named.
• Store notes safely. Keep them in a shared team folder, not in public chats. Record only with consent, and delete recordings once you no longer need them.
• Respect confidentiality. If a founder tells you something "off the record," it stays off the record.
• Thailand has a personal data law, the Personal Data Protection Act (PDPA), fully in force since 1 June 2022². You do not need to be a lawyer for a student interview — but the spirit is simple: collect only what you need, say why, and keep it safe.

If you interview as a pair or a team, write one log together. And make sure everyone named in it was really there.

How do you turn your logs into an interview report?

This is the order I follow.

1. Line them up. Put all logs side by side, by type.
2. Count the facts, not the compliments. Which facts and behaviors appear in more than one interview?
3. Find the contradictions. Are they really disagreement — or two different segments?
4. Name what changed. For each change in your idea, link the interviews that caused it.
5. Say what you still do not know. The open questions are your plan for the next stage.

Frequently asked questions

What counts as evidence in a customer interview?
What happened (facts), what people already did (past behavior), and what they gave up for you (time, reputation or money). Opinion is what people think, predict or would do. Only evidence should drive decisions.

Is "I would buy that" evidence of demand?
No. It is a hypothetical. A pre-order, a deposit, a second meeting or an introduction is evidence.

Do you need consent to record a customer interview?
Yes. Ask clearly before you start, record only after a yes, and delete the recording once you no longer need it. Many people speak more freely without a recorder, so good notes written the same day are often better.

Can I use AI to write my interview notes?
I recommend against it. Notes in your own words keep the surprise. Use AI for research and drafting elsewhere, and be open about what it helped with.

✦ A Note for Global Venture I Students · Block C: Understanding the Market · Form C

Every one of your twelve interviews needs a log entry using the fields above, and a line in your Venture Log. Entrepreneur interviews use the Entrepreneur Interview form, customer interviews the Customer Interview form, and both friend types the Friend Interview form. All of them feed your interview report, one of the five final deliverables. In Form C: entrepreneur interviews #2–3 and customer interviews #1–3.

Interview rules: each interview counts once, as one type. Entrepreneurs and customers must be strangers to you; an introduction from family is fine. Group interviews are allowed, and online interviews count. If you interview as a pair or team, write one log together and credit is divided by group size (two of you = half an interview each) — so make sure everyone who gets credit was really there.

Thank-you emails and proof: within 24 hours, with kasper@seabridge.space in CC (not BCC) — one from each group member who took part. Proof of each interview is a selfie, with permission; if someone declines a photo, a screenshot of your thank-you email is the evidence.

AI rule: AI is welcome for research and drafting, but your interview notes, Venture updates, reflections and letters must be your own words, and every form has an "AI used for:" field where you say what AI helped with.

Evidence bars: every block form and the interview report are read against the A / B / C / Not yet bars above. Your KTH readiness self-score works the same way: a level only counts when you can point to the evidence. An honest lower score is never penalized.

Storage: keep notes in your team's shared course folder. Record only with consent, and delete recordings after the course. The open questions at the end of your interview report become your plan for Global Venture II: Sprint.

Bring to the debrief: one completed log entry with each note tagged as fact, behavior, commitment or opinion.

Do this before class:

1. Copy the log fields into your team's shared folder as a template. Fill it for every interview you have done so far.
2. Go back through your notes and tag each line: F (fact), B (past behavior), C (commitment), O (opinion). Count how many are O.
3. In your Venture Log, add one line per interview: date · type · biggest surprise · link to the entry.

Pre-class check (in Form C):

1. Name the three kinds of evidence, with an example of each from your own interviews.
2. Rewrite one opinion from your notes as a question that would have produced a fact instead.
3. What will you do if someone does not want to be photographed?

Venture update (Form C): tool applied · what changed in the idea · evidence link · readiness self-score. Each member writes their own update, in their own words (your team keeps one Venture Log), and fills in the form's "AI used for:" field.

Your progress: Customer readiness to level 3, stretch 4 (the problem confirmed by several customers, with evidence you can show). How we measure progress

Skills you practice: Empathy and active listening · analytical thinking · spotting opportunities · ethical and sustainable thinking.

Keep reading

• The Interview Masterclass — how to get facts, not compliments
• First 10 Customers & the Secret of Angel Customer™ — next in Block C: who to log next
• The Venture Log & Book Log — where every log entry is linked
• The KTH Innovation Readiness Level: How We Use It (and What We Expect) — scoring with evidence

All articles in Block C · Understanding the Market

• C-01 Market Research for Startups
• C-02 Choose Your Beachhead
• C-03 The Interview Masterclass
• C-04 How Many Customer Interviews Are Enough?
• C-05 The Interview Log & Evidence Rules (you are here)
• C-06 First 10 Customers & Angel Customer™
• C-07 Jobs-to-Be-Done
• C-08 The Competitive Reality Map™
• C-09 Double Diamond & Design Thinking

References

1. Fitzpatrick, R. (2013). The Mom Test: How to talk to customers and learn if your business is a good idea when everyone is lying to you. Robfitz Ltd. https://www.momtestbook.com/ (quoted line: https://www.goodreads.com/quotes/11029920-commitment-can-be-cash-but-doesn-t-have-to-be-think; summary of the rules: https://mtlynch.io/book-reports/the-mom-test/) — accessed Nov 2026.
2. Tilleke & Gibbins. (2022, June 30). First set of subordinate regulations enacted for Thailand's PDPA. https://www.tilleke.com/insights/first-set-of-subordinate-regulations-enacted-for-thailands-pdpa — accessed Nov 2026.

Further reading

• Fitzpatrick, R. (2013). The Mom Test: How to talk to customers and learn if your business is a good idea when everyone is lying to you. CreateSpace.
• Blank, S., & Dorf, B. (2012). The startup owner's manual: The step-by-step guide for building a great company. K&S Ranch.

About the author: Kasper-Tanakrit Sermsuksan is the Founder of SEA Bridge and a Visiting Lecturer in Computer Engineering and Digital Technology at Chulalongkorn University's Faculty of Engineering. Learn more →