Making the go / pivot / kill call with evidence

Why this matters

The hardest founder decision is "should I keep going?" Most say yes because they've already invested time. That's the sunk-cost trap. A good decision framework looks at the evidence ONLY — assumptions supported vs. invalidated, signal strength, market size — and gives you a recommendation you can defend to a co-founder, investor, or yourself three months from now. Even if you disagree with the output, seeing it written down changes how you think.

What you'll do in this step

• Margawise scores your evidence: critical assumptions validated (weighted 40%), high assumptions (25%), interview count, experiment outcomes, GTM signal, market size.
• It produces a confidence percentage from 0-100.
• Based on the score, one of 4 recommendations: Build MVP (≥72), Experiment More (≥50), Archive (≥30), or Kill (<30).
• You can accept the recommendation, pick a different one with your own reasoning, or go back and collect more evidence if you're not ready.

Common mistakes to avoid

• Trusting the score blindly — it's a summary of YOUR data. If your interviews were bad, your score is bad too.
• Treating "Build MVP" as permission to stop validating — building itself is a continuous validation.
• Killing too quickly — if confidence is borderline, "Experiment More" is usually the right call. Founders who kill at 45% often should've run 2 more experiments.
• Refusing to pivot — pivoting based on evidence is a strength, not a failure. Slack, Instagram, and Twitter all pivoted.

How Margawise helps

• Score is computed from every stage's data — not a guess, a weighted sum.
• Commit threshold of 72 matches the public benchmark on our homepage so you know what "ready" looks like.
• AI generates the rationale in plain English — "your willingness-to-pay assumption didn't hold; consider alternative monetisation."
• If you accept the recommendation, Margawise unlocks follow-up tools: PRD generation for Build, GTM plan, or an archive reminder for a future return.