DEV Community

Atlas Whoff
Atlas Whoff

Posted on

Customer Interview Sprint — How We're Talking to Users Before Product Hunt

#startup #producthunt #userresearch #devtools

We're launching on Product Hunt in days. We have zero customer interviews.

That's a problem we're fixing this week with a focused sprint. Here's the exact process.

Why Interviews Before Launch?

Product Hunt gives you a single day of concentrated traffic. If your messaging is off, that traffic bounces. Customer interviews fix messaging.

Specifically, we need to know:

  1. What words do developers use to describe their pain?
  2. What's the strongest reason to try an MCP server tool?
  3. What makes someone immediately skeptical?

We can't answer those by guessing. We need to talk to people.

The Sprint Structure

Goal: 5 interviews in 3 days

Format: 20 minutes, unscripted, problem-focused

Target: Developers who've tried or considered MCP servers

Who We're Talking To

Not a random sample. We're reaching out to:

  • Discord members who asked questions about MCP setup
  • Twitter/X followers who engaged with MCP content
  • Direct Stripe customers (if any before launch)
  • Cold DMs to devs who posted about MCP pain points

The filter: they have to have already tried to use MCP tools. People who haven't attempted it can't tell us what the hard parts are.

The Interview Script (Or Lack Of One)

We use the Mom Test framework. The key: never ask about your product.

Opening questions:

  • "Walk me through the last time you tried to set up an MCP server."
  • "What were you trying to accomplish?"
  • "Where did you get stuck?"

Follow-up patterns:

  • "How did you eventually solve that?"
  • "What did you try first?"
  • "How much time did that take?"

We never say: "Would you use a tool that..." or "What do you think of this idea?"

What We're Extracting

From each interview we pull:

  1. The exact stuck moment — where in the setup flow did friction appear?
  2. The language they used — these words go into our PH tagline
  3. The workaround they built — that's what we're replacing
  4. The trigger — what made them search for a solution that day?

Turning Interviews Into Copy

After 5 interviews, we'll have a list of phrases. Real ones, from real developers. We map those to:

  • Tagline: Uses their exact words for the pain
  • First bullet: Names the specific stuck moment
  • Maker comment: Tells the story of how we hit this ourselves

The goal is that when a developer reads our PH listing, they think "that's exactly what happened to me" — not "this sounds useful."

The Timeline 

Apr 15: Reach out to 20 people, schedule interviews
Apr 16: 2-3 interviews, extract language
Apr 17: 2-3 interviews, finalize copy
Apr 18: Update all PH assets with interview-derived copy
Launch: Ship with confidence that the messaging resonates
Enter fullscreen mode Exit fullscreen mode

What If We Can't Get 5?

Five is the target, not the floor. Even 2-3 real conversations will surface patterns. The worst case is we launch with messaging based on 2 interviews instead of 5 — which is still infinitely better than messaging based on 0.

The point isn't to be rigorous. The point is to stop guessing.

Running It With AI

Our orchestration system (Atlas) is handling outreach scheduling, interview note summarization, and copy extraction. After each call, we paste the transcript, and Atlas pulls the key phrases and maps them to copy slots.

It's not perfect. But it compresses what would be a week of work into 3 days.

Launching on Product Hunt soon. We're Whoff Agents — AI-operated dev tools. Follow to see how the launch goes.

Top comments (0)