How to Do Customer Interviews for Messaging in Crypto

To do customer interviews for messaging, talk to 8 to 12 real users one at a time, ask them to describe the problem and the product in their own words, and record the exact phrases they use. Then count which phrases repeat across calls and put the most common ones into your homepage, pitch and press materials. You're not testing features. You're collecting the words your market already uses so your copy stops sounding like your engineering team.

Most crypto teams write messaging from the inside out. The founder knows the architecture, so the homepage describes the architecture. Interviews flip the direction. The best line on your website is often something a user said on a call last Tuesday.

Why interviews fix messaging faster than brainstorming

A messaging workshop with your own team produces the language your team already uses. That's the problem you were trying to solve.

Users describe the product by what changed for them. They say "I stopped losing money to bridge fees" where your deck says "unified liquidity layer". They say "my mum could do this" where your site says "abstracted UX". The user version is almost always plainer, more concrete and easier for a journalist to repeat.

I've been saying a version of this since 2022, when I was thinking about retention rather than copy:

you should be focusing on user expectations time & again & investing on user survey every quarter to align expectations. Shilika Jain on X, 6 Dec 2022

Surveys tell you what people expect. Interviews tell you how they say it. For messaging you need the second one, because the exact wording is the asset.

If you haven't yet diagnosed whether your problem is the product or the explanation, start with my piece on the explanation problem in crypto. Interviews are the repair kit once you know which one you have.

Who should you interview for crypto messaging research?

Not just your fans. A good interview pool has three groups, because each one tells you something different about your words.

Aim for roughly half stayers and split the rest between leavers and lookers. Stayers give you the phrases to use. Leavers and lookers show you the phrases to delete.

One filter matters more in crypto than anywhere else: screen out people who only showed up for points or an airdrop. They'll describe the reward, not the product, and their language will pull your copy towards incentives. I wrote about that gap in airdrops versus real users. For messaging research, the farmers are noise.

How many customer interviews do you need?

Start with 8 and stop when the phrases start repeating. For most early-stage protocols and apps that happens somewhere between call 8 and call 12.

The logic is simple. You're not measuring a percentage, you're looking for saturation: the point where a new call gives you no new way of describing the problem. If call 11 sounds like calls 4, 7 and 9, you're done. If every call surprises you, you either have several distinct user types or your product means different things to different people, and that's a positioning finding in itself.

A few rules of thumb I use:

• If you serve two clearly different audiences (say, developers and end users), run a separate set of 6 to 8 for each. Don't blend them.
• Run calls over about two weeks, not two months. Your product changes and older calls go stale.
• Keep calls to 30 minutes. Longer calls drift into feature requests.

What questions should you ask in a messaging interview?

Ask about their life before the product, the moment they decided to try it, and how they'd describe it to someone else. Don't ask what they think of your tagline. People are polite about taglines and honest about their own problems.

Here's the question list I give founders. Use it in this order.

1. What were you trying to do when you first found us?
2. What were you using before, and what annoyed you about it?
3. Walk me through the first time you used the product. What did you expect to happen?
4. Was there a moment you almost gave up? What was confusing?
5. If a friend outside crypto asked what this is, what would you tell them?
6. What would you lose if this disappeared tomorrow?
7. Who else do you think needs this?
8. Is there a word we use on the site that you didn't understand or ignored?

Question 5 is the one that pays for the whole exercise. Write down the answer word for word. Don't tidy it up and don't finish their sentences when they pause. The pauses tell you where the explanation is missing.

Question 8 gives you a free jargon audit. If three people independently mention the same term, it goes on your cut list. My companion piece on crypto jargon to cut from marketing has replacements for the usual suspects.

How do you turn interview phrases into copy?

This is where most teams stop short. They do the calls, feel enlightened, and then write the homepage from memory. Memory edits. Use a log instead.

I call it the phrase harvest. After each call, go through the recording or transcript and pull every phrase where the user describes a problem, an outcome or the product itself. Put each one in a sheet with four columns.

The example row is invented to show the format, so don't read it as a real user quote.

When the sheet is full, sort by count. The top three phrases are your raw material. You'll rarely paste them in as-is, but you'll keep their nouns and verbs. "Send money home" beats "cross-border settlement" every time it appears in a real mouth.

Then route them:

• Homepage hero: the top outcome phrase from stayers, lightly edited.
• Pitch angle for media: the problem phrase that shows up across all three groups. That's the tension a reporter can build a story on.
• FAQ and onboarding: the confusion phrases from leavers, answered directly.
• Delete list: any term of yours that lookers misread or skipped.

If you're rewriting the homepage off the back of this, the five-second homepage test is the next step.

How does this feed into PR and content?

The phrases that survive the harvest become the spine of everything outward-facing, which is why I run this before a launch campaign rather than after it.

Journalists are pitched all day with feature lists. A pitch built on a problem real users describe in plain words reads differently, because it's already a story about people. I tell founders to treat the problem phrases as a content calendar too:

List potential issues with products and services and create problem solving content around it! Shilika Jain on X, 30 Jan 2023

Every confusion point a leaver raised is a post, a thread or an FAQ entry waiting to be written. Every outcome a stayer described is a candidate angle for an op-ed or a founder interview. In my Web3 PR campaigns this research usually sits in week one, before a single pitch goes out, because it decides which story we tell.

Common mistakes in customer interviews for messaging

A short list, from seeing teams try this:

• Pitching during the interview. If you explain the product in the first five minutes, you've contaminated every answer after it. Let them explain it to you.
• Only calling Discord regulars. Your most active community members already speak your language. They're the least useful people to learn plain language from.
• Paying in tokens. A token incentive tells the interviewee what you want them to care about. If you must thank people, keep it small and unrelated to the product's economy.
• Summarising instead of quoting. "Users find fees confusing" is a summary. "I had no idea why it charged me twice" is a phrase you can use.
• Letting the founder run every call. Founders defend. Have someone else run at least half of them.

When customer interviews won't help

Two situations where I'd hold off.

If you have no real users yet, interviewing prospects about a product they've never touched gives you opinions, not experience. Use it to understand the problem, but don't expect copy-ready phrases.

If your users are deeply crypto-native and anonymous, many won't take a video call. Async works better there: a short voice note prompt in a Telegram DM, or three written questions. You lose some nuance and gain a much bigger sample. That's a fair trade.

And if stayers can describe the product clearly but still churn, the problem isn't your messaging. No amount of better copy fixes a product people understand and don't want.

FAQ

Should I record customer interviews?

Yes, with permission. You need the verbatim wording later and nobody types fast enough to catch it live. Tell the interviewee the recording is only for your team's notes and won't be published.

Can I use AI to transcribe and summarise the calls?

Transcription, yes. Summaries are risky for this specific job because they paraphrase, and paraphrase is exactly what you're trying to avoid. Read the transcript yourself and pull the phrases by hand.

How often should I repeat messaging interviews?

Run a fresh round whenever your audience shifts: a new chain, a new region, a move from developers to consumers. Otherwise a light round every quarter keeps your language from drifting back towards internal jargon.

Do I need a researcher to do this?

No. A founder or marketer can run these with the question list above. What you do need is someone willing to hear that the homepage they wrote isn't working.

If you've got interview notes and still can't land the sentence, book 30 minutes with me and we'll pull the hero line out of them together.