Designing a “Friction-Free Experiment” experience in Heatseeker
Introduction
83 % of growth teams who create a Heatseeker trial account never launch their first experiment.ⁱ For a product that promises “real-market evidence in minutes,” that drop-off was mission-critical.
Design brief
How might we help a first-time user sign in, glide through onboarding, and launch their very first experiment in under 90 seconds?
Scope → Sign-in · Onboarding · Dashboard · “Start Experiment” flow
Context
Heatseeker’s secret sauce is a blend of stealth ads (to gather live demand data) and synthetic personas (to predict scale). Powerful, but also intimidating for new users who just want a “quick A/B test.”
Persona
Our core persona for this project is, let’s call them “Launch-Hungry Sara” is a solo growth PM:
Proficient in English UX patterns
Very comfortable with SaaS dashboards
Ultra-impatient: bounce if value isn’t obvious in < 2 minutes
The legacy experience required Sara to :
Sign in (SSO or email & password only)
Choose the thing she is testing, Audience & Goal
Generate & Review Experiment
Set the Budget
Simulate / Launch
Design Exploration
Option 1 (Safe Option)
Goal:
Just Clean It Up User Problem: “I’m confused. Where do I look? There’s too much info and no clear start.”
User Journey:
Lands on dashboard, sees messy info.
Can’t find what matters, too many elements shouting at once.
Doesn’t know what to do next → bounces.
Essential (Safe Option)
Option 2 (Stretch Option)
Goal:
Make it Goal-Focused User Problem: “I don’t know what my goal is. Am I testing the right things?”
User Journey:
Lands on dashboard, sees lots of data.
Feels disconnected from why they’re running experiments.
Unsure which metric matters most for their growth.
Goal-Centric (Stretch Option)
Option 3 (Wild Option)
Goal:
AI Growth Copilot Mode. User Problem: “I don’t know what to test next. I just want results fast.”
User Journey:
Logs in.
Sees a list of experiments and a graph, no context.
Thinks: “What should I do now?”
Leaves or starts a random test without much strategy.
AI-Powered Assistant (Wild Option)
Verdict
The Wild option not only meets but comfortably outperforms the client’s
“< 90 sec to launch” requirement, with every tester completing in under 1 min 30 sec and strong confidence scores to match.