Pivot, persevere, or kill

After the learning vehicle is live, look for evidence — not vibes. Decide in advance what would cause a pivot, persevere, or kill. Kill is underrated: stopping a false path is a successful MVP. Persevere means deepen the wedge (better reliability, adjacent jobs) — not instantly expand to every persona on the vision slide.

Signals to watch

Walking through meeting notes

You ran a concierge MVP for "AI meeting notes for sales calls" with 12 design partners for four weeks. Pre-declared rules:

• Persevere if ≥6 pay a pilot invoice and ≥8 reopen notes weekly without you nudging.
• Pivot if they love notes but only when you manually fix CRM fields — maybe the job is CRM hygiene, not transcription.
• Kill if fewer than three pay and interviews show the pain is "nice to have" behind a real priority.

Hitting kill criteria is not embarrassment. It is cheap insurance against a nine-month build.

Four lines before any "MVP" project

Write: (1) the assumption under test, (2) the smallest experience that can falsify it, (3) the metric or qualitative evidence you will accept, (4) the appetite in weeks. If you cannot write those four lines, you are not scoping an MVP — you are starting a build.

Put those four lines at the top of the PRD or pitch. When someone asks you to "design an MVP for X," name assumptions, pick a type, define success/fail criteria, then describe what you would build second if the signal is good. Say when you would not use an MVP framing because the risk is already resolved.

Common mistakes

• Deciding from vibes. Pre-declare numbers. "They seemed excited" is not a signal.
• Persevere by expanding the vision. Deepen the wedge first.
• Treating kill as failure. A clean kill is the MVP working.

Check yourself

Six of 12 partners pay; eight reopen weekly. Pivot, persevere, or kill? Persevere. Both pre-declared bars cleared. Deepen the wedge; do not add every persona.

They love the notes only after you fix CRM fields by hand. What now? Pivot. The job may be CRM hygiene, not transcription. The concierge revealed it.

Key takeaways

• Predefine pivot / persevere / kill; treat a clean kill as success.
• Watch retention, payment, pull, specific complaints and organic sharing — not compliments.
• Four lines on the pitch: assumption, smallest test, evidence, appetite in weeks.

Further learning

Eric Ries at Talks at Google on Lean Startup, validated learning and MVPs:

SVPG articlesA partnership dedicated to teaching best practices to product teams and product leadersSilicon Valley Product Group

Map: MVPs and scoping.