Experiment Types
Select MVP experiment formats based on risk and evidence strength.
Key takeaways
- Choose the experiment format that attacks the riskiest assumption first, because each type answers a different question.
- The menu spans landing pages, concierge MVPs, clickable prototypes, thin product slices, paid pilots, and community/content tests.
- Match the main risk to the format: message risk points to a landing page, willingness-to-pay risk points to a paid pilot.
- Agents make multiple test surfaces cheap, so use that for controlled comparison rather than feature sprawl, keeping one primary question.
Different experiments answer different questions. Choose the format that attacks the riskiest assumption first.
Experiment Menu
| Type | Best for | Output |
|---|---|---|
| Landing page | Offer and channel validation | Conversion and message signal |
| Concierge MVP | Workflow and value validation | Manual delivery evidence |
| Clickable prototype | UX and information architecture | Task completion and objections |
| Thin product slice | Activation and retention | Usage and support data |
| Paid pilot | Willingness to pay | Contract, budget, and success criteria |
| Community/content test | Audience and problem language | Engagement and response patterns |
Selection Guide
| Main risk | Start with |
|---|---|
| We do not know if the problem matters | Interviews or concierge MVP |
| We do not know if the message works | Landing page |
| We do not know if users understand the flow | Prototype |
| We do not know if the product can deliver value | Thin product slice |
| We do not know if customers will pay | Paid pilot |
Agentic Advantage
AI agents make it practical to build several test surfaces quickly. Use that advantage for controlled comparison, not feature sprawl. The strongest experiments still have one primary question.