Signal Over Metrics (Early Stage)

"You can't optimize your way to product-market fit. I don't care at the early stages if something's optimized by 5% from an email. That doesn't fundamentally tell me if something's working or not." - Claire Butler

What It Is

A framework for decision-making in early-stage companies that prioritizes qualitative signals over quantitative optimization. When user numbers are small, statistical significance is impossible, and metric improvements can be misleading. What matters is finding people who truly love the product—even just a few of them.

This doesn't mean ignoring data entirely. It means understanding that the right question at early stages isn't "how do we optimize this metric?" but "can we find anyone who genuinely loves this?"

The key signals Claire Butler looked for:

• Users who wanted to "take the laptop out of Dylan's hands" because they were so excited
• Teams willing to use Figma full-time before it was even ready
• Emotional reactions during demos, not just polite interest
• People willing to stake their reputation on recommending the product

How It Works

Why Metrics Fail Early:

1. Sample Size - With 10 users, a 5% improvement means half a person. Statistically meaningless.

2. Noise Dominance - At small scale, random variation swamps any real signal in the data.

3. Optimization ≠ PMF - You can A/B test your way to a slightly better landing page while building something nobody wants.

4. False Confidence - Improving a metric feels like progress even when you're optimizing the wrong thing.

What Signals Tell You:

• Love vs. Like - Do users merely use your product, or do they actively champion it?
• Organic Pull - Are users asking to use more, or do you have to convince them?
• Emotional Investment - Do users care when things break, or do they just leave?
• Willingness to Switch - Will users go through the pain of changing their workflow?

The Progression:

Early stage: Can you find ONE person who loves it? Next: Can you get them to KEEP using it? Then: Can you find a SECOND person who loves it? Later: Can you get someone to PAY for it?

These are sequential milestones, not parallel metrics to optimize.

How to Apply It

1. Define Your Signals

   • What would a user who truly loves your product do?
   • What emotions would they display?
   • What actions would they take unprompted?

2. Gather Signals Actively

   • Demo to users in person when possible
   • Watch their reactions, not just their words
   • Note the difference between polite interest and genuine excitement

3. Look for Pull

   • Are users pulling the product from you, or are you pushing it on them?
   • Do they reach out proactively, or only respond when contacted?
   • Do they tell others without being asked?

4. Trust Your Intuition

   • Build confidence in reading signals
   • Develop pattern recognition for genuine vs. performative interest
   • Accept that this requires judgment, not just data

5. Use Metrics Later

   • Once you have enough scale for statistical significance
   • Once you've validated that people love the core product
   • For optimization, not validation

When to Use It

Apply this framework when:

• You have fewer than hundreds of active users
• You're still searching for product-market fit
• You're trying to validate core value propositions
• A/B tests would lack statistical power

Shift to metrics when:

• User numbers reach statistical significance
• You've validated genuine love exists
• The question shifts from "is this working?" to "how do we improve?"
• Optimization becomes more valuable than validation

Source

• Guest: Claire Butler
• Episode: "An inside look at Figma's unique GTM motion"
• Key Discussion: (00:23:19) - Why signal beats metrics early on
• YouTube: Watch on YouTube

Related Frameworks

• Minimum Lovable Product - Build products people love, not just products that work
• Reference Customer Development - Find customers willing to stake their reputation on you