Lean Validation Techniques
Purpose: choose the lightest validation method that can confirm demand, usability, or value before full implementation.
Contents
- Method comparison
- Selection rules
- Fake Door
- Wizard of Oz and Concierge
- Lightweight PRD, RFC, ADR
- Spec-Driven Development
Pre-Build Validation Methods
| Method | Cost | Typical duration | Best for | Fidelity |
|---|---|---|---|---|
Fake Door Test |
very low | hours to 1 day |
demand and interest | low |
Wizard of Oz |
low to medium | 1-2 weeks |
user experience and usability | medium to high |
Concierge MVP |
medium | 2-4 weeks |
value and workflow | high |
Prototype (Forge) |
medium | 1-2 weeks |
UX and core flow | medium |
A/B Test (Experiment) |
high | 2-6 weeks |
measured effect on outcomes | high |
Selection rules:
- unknown demand ->
Fake Door Test - demand seems real but usability is unclear ->
Wizard of OzorConcierge MVP - solution options need comparison -> prototype or
A/B Test
## Pre-validation Plan
Use this block in proposals that need early validation.
Required fields:
- validation method
- placement or entry point
- target segment
- success metric
- sample size
- what to do if the result misses the threshold
Fake Door Test
Use patterns such as:
- CTA button to a waitlist
- menu item to a "Coming Soon" state
- in-app invite to a beta
- pre-order or early-access signup
Advanced metrics to watch:
- unique vs total clicks
- time to click
- post-click continuation vs bounce
Suggested proposal fields:
PlacementCTA TextSuccess Criteriasuch asCTR >= ___%DurationSample Size- below-threshold action: reconsider or pivot
Wizard of Oz And Concierge MVP
| Method | User perception | Best for | Main risk |
|---|---|---|---|
Wizard of Oz |
users think the system is automated | UX and experience validation | ethical transparency and operator cost |
Concierge MVP |
users know humans provide the service | value and workflow validation | very limited scale |
Use Wizard of Oz when:
- the feature is technically complex
- experience quality matters more than scalability
- you need pre-build learning for AI-like behavior
Use Concierge MVP when:
- the full workflow must be validated end to end
- a human-delivered service can stand in for automation
Lightweight Spec Formats
| Format | Best use | Notes |
|---|---|---|
| Lightweight PRD | quick alignment | concise and iteration-friendly |
1-Pager |
fast stakeholder alignment | compresses key detail into one page |
Amazon 6-Pager |
narrative-heavy strategic framing | dense reading, no slideware |
RFC + ADR |
changes needing feedback and a recorded decision | keep ADRs short |
RFC/ADR rules:
- use
RFCwhen the change affects others and needs buy-in - use
ADRto record the chosen implementation direction after review - keep the RFC review window to
2-3 days - keep the process lightweight
Spec-Driven Development
Treat the proposal as an input artifact for AI-assisted build systems.
Implications for Spark:
- structured specs improve downstream build quality
- technical constraints must be explicit
- Builder handoffs should be ready for spec-first execution
Reference tool styles:
- Amazon Kiro: requirements -> design -> tasks
- GitHub spec-kit: multi-step spec plus constitution
- Tessl: specification as a maintained artifact
Spark Phase Mapping
| Spark phase | Validation technique | Purpose |
|---|---|---|
IGNITE |
discovery cadence | continuous opportunity discovery |
SYNTHESIZE |
ODI opportunity scoring | compare candidate opportunities |
SPECIFY |
lightweight PRD or RFC | document the idea quickly |
VERIFY |
Fake Door, Wizard of Oz, Concierge | validate before build |
PRESENT |
1-Pager or SDD-ready spec | align stakeholders and downstream agents |