← Blog

The Evidence Sprint: Prove Value in 90 Seconds

7 min readNicolas Lester Reynolds
#evidence-sprint#rapid-prototyping#mvp

Most MVPs fail before they launch.

Not because the idea is bad. Not because the team is incompetent. But because it takes 3 months to build and 3 seconds to realize it's not what the client needed.

The Evidence Sprint solves this: working demo + quantified delta in 2-5 days.

The Problem with Traditional MVPs

The term "MVP" has lost all meaning. Here's what it's become:

"Minimum Viable Product" in practice:

  • 12 weeks of development
  • 47 features (because "we need X to be viable")
  • Launch day: clients try it, realize it's not what they wanted
  • Total waste: 12 weeks, $50K-$150K
The core issue: You don't learn until you ship. And shipping takes months.

Why MVPs Take So Long

1. Feature Creep

  • "We need user auth to be viable"
  • "We need notifications to be viable"
  • "We need analytics to be viable"
Result: 6-month "MVP"

2. No Forcing Function

  • If you have 3 months, you'll use 3 months
  • Parkinson's Law: work expands to fill time
  • No pressure to prioritize ruthlessly
Result: Unfocused work

3. No Quantified Success Criteria

  • "Let's build it and see if users like it"
  • How do you measure "like"?
  • When do you know if it's working?
Result: Ambiguous outcomes

The Evidence Sprint Approach

Instead of building a minimum viable product, we build a minimum viable proof.

1

Working Demo (≤90 seconds)

Not a mockup. Not a prototype. A working implementation you can use.

The 90-second rule forces clarity:
  • If you can't demo it in 90 seconds, is it focused enough?
  • If the value isn't obvious in 90 seconds, will users stick around?
2

Quantified Delta

Not "it's faster" but how much faster.

Not "it's simpler" but how many fewer steps.

Every Evidence Sprint produces DELTA.md with before/after numbers.
3

Tag + Proof Entry

Tag the work: evidence-sprint_<milestone>_<date>

Generate public proof entry with:
  • AC.md (what we're proving)
  • DEMO.md (where to see it working)
  • DELTA.md (quantified improvements)

Real Example: OTP Signup

Client need: Passwordless authentication, fast.

Traditional MVP approach:

Week 1-2Requirements gathering, architecture design
Week 3-6Build complete auth system
Week 7-8Polish and testing
Week 9Ship and hope it works
Total: 9 weeks, $40K-$80K

Evidence Sprint approach:

Day 1

Co-write AC.md

# AC.md: OTP Signup Evidence Sprint

## Functional Criteria
1. User enters email, receives OTP
2. User submits OTP, authenticates
3. Session persists 30 days

## Non-Functional Criteria
- p95 latency: < 300ms (full flow)
- Steps to complete: ≤ 3
- Mobile responsive

## Verification
npm run acceptance:signup
Days 2-3

Build minimal proof

  • Email OTP delivery (no SMS, no backup codes, no recovery)
  • Simple session management (no OAuth, no SSO)
  • Basic error handling (no edge cases yet)
Day 4

Generate delta

MetricBeforeAfterChange
p95 latency1200ms280ms↓77%
Steps to complete73↓57%
Forgot password flow3 stepseliminated-
Mobile conversion42%68%↑62%
Day 5

Ship demo + tag

Demo:demo.scopelock.ai/otp-signup
Tag:evidence-sprint_otp-signup_2025-11-02
Proof:/proof/evidence-sprint_otp-signup_2025-11-02
Total: 5 days, $3K-$6K

The 90-Second Demo Rule

Why ≤90 seconds?

1. Forces Focus

If you can't show value in 90 seconds, you're building too much.

✓ Good Evidence Sprint demo:

Open page → Enter email → Receive OTP → Submit → Authenticated

Time: 45 secondsValue: Obvious
✗ Bad Evidence Sprint demo:

Explain architecture → Show database schema → Walk through 5 features

Time: 8 minutesValue: Unclear

2. Matches Real User Attention

Users don't give you 10 minutes to "show potential."

They give you 30 seconds to prove value or they bounce.

3. Enables Decision-Making

In 90 seconds, a client can decide:

"This is exactly what I need—let's build to AC green"
"Close, but change X—let's iterate"
"Not what I wanted—let's pivot before investing more"

All outcomes are valuable. Finding out in 5 days vs 5 months is the difference.

Quantified Deltas: Not Feelings, Numbers

Every DELTA.md must have measurable before/after.

❌ Bad Delta

## What Changed
- Made it faster
- Simplified the flow
- Improved UX
Why bad? No numbers. Can't verify. Just claims.

✅ Good Delta

## Performance
- p95 latency: 1200ms → 280ms (↓77%)
- Time to first interaction: 850ms → 190ms (↓78%)

## Usability
- Steps to complete: 7 → 3 (↓57%)
- Form fields: 12 → 2 (↓83%)

## Conversion
- Mobile signup rate: 42% → 68% (↑62%)
- Completion rate: 54% → 81% (↑50%)
Why good? Every claim is a number. Client can verify. Objective proof.

Evidence Sprint vs Discovery Phase

Most agencies sell a "discovery phase" before building. Here's the difference:

AspectDiscovery PhaseEvidence Sprint
DeliverableDocumentWorking demo
Timeline2-4 weeks2-5 days
Output"Here's what we'll build""Here's what we built"
Cost$5K-$15K$3K-$6K
RiskStill don't know if it worksYou've seen it working
Discovery Phase: Pay to talk about building.
Evidence Sprint: Pay to build and prove.

When to Use Evidence Sprint

✅ Perfect For

  • Validating new ideas: "Will OTP signup work for our users?"
  • De-risking big projects: "Can we hit <300ms p95 before committing?"
  • Rapid prototyping: "We need a working demo for investors by Friday"
  • Proof before commitment: "Show me it works before I approve the full build"

❌ Not Ideal For

  • Well-defined features: If you know exactly what you want and have full specs, go straight to AC green
  • Pure research: If you're exploring problem space without building, use a different approach
  • Long-term maintenance: Evidence Sprints prove value, not operational readiness

How to Apply This

If You're a Founder

Before committing to a 3-month build, ask your developer:

"Can we do an Evidence Sprint first?"
What you should get:
  • Working demo in 2-5 days
  • Quantified delta (numbers, not feelings)
  • Clear path: iterate, proceed to AC green, or pivot
Cost:$3K-$6K typically (1/10th of full build cost)
Value:Knowing if you're building the right thing before investing $50K+

If You're a Developer

Instead of quoting 8 weeks for an MVP:

"Let's do a 3-day Evidence Sprint first."
What you deliver:
  • Minimal proof of concept (working, not polished)
  • DELTA.md with real numbers
  • Demo link clients can share with stakeholders
Why clients say yes: Low risk, fast feedback, proof before commitment.

Three Common Questions

"Isn't this just a prototype?"

No. Prototypes are fake (Figma mockups, clickable demos).

Evidence Sprints are working implementations:

  • Real database
  • Real auth
  • Real performance metrics
  • Real code that becomes production

"What if the Evidence Sprint proves it won't work?"

That's success.

You spent 5 days and $3K-$6K to learn this won't work.

Alternative: Spend 12 weeks and $80K to learn the same thing.

"Do you always do Evidence Sprints?"

No. If requirements are crystal clear and risk is low, we skip straight to AC green.

Evidence Sprints are for validation and de-risking, not for every build.

The Math That Matters

Traditional MVP

Time to learn:12 weeks
Cost to learn:$50K-$150K
Pivots:Expensive and slow

Evidence Sprint

Time to learn:2-5 days
Cost to learn:$3K-$6K
Pivots:Cheap and fast

The difference:

24xfaster feedback
10xlower cost
more pivots possible

What's Next?

At ScopeLock, we start high-risk projects with Evidence Sprints.

1Co-write AC.md (30-60 minutes)
2Build Evidence Sprint (2-5 days)
3Review delta + demo (30 minutes)
4Decide: iterate, proceed to AC green, or pivot
$3K-$6K, working demo + quantified delta