# CRO Experiments Library

Proven A/B test ideas by page type.

## Homepage Tests

**Headline:**
- Outcome-focused vs. feature-focused
- Specific number vs. generic
- Question vs. statement

**Hero CTA:**
- "Start Free Trial" vs. "Get Demo"
- Button color (high contrast wins)
- Above fold vs. below fold

**Social Proof:**
- Customer logos vs. testimonial
- User count vs. case study snippet
- Placement: Hero vs. mid-page

## Landing Page Tests

**Form Length:**
- Email only vs. Email + Name
- Short form vs. Long form (qualification)

**Value Proposition:**
- Lead with benefit vs. lead with how it works
- Video vs. text + image

**Trust Signals:**
- Security badge vs. money-back guarantee
- Social proof type (logos vs. testimonials vs. stats)

## Pricing Page Tests

**Pricing Display:**
- Monthly vs. Annual default
- 3 tiers vs. 4 tiers
- Highlight most popular

**Feature List:**
- All features vs. top 5 per tier
- Icons vs. checkmarks

**CTA:**
- "Start Free Trial" vs. "Choose Plan"
- Single CTA vs. CTA per tier

## Signup Flow Tests

**Fields:**
- Email only → password later
- Name + Email vs. Email only
- Social signup (Google/GitHub) vs. manual

**Validation:**
- Inline validation vs. submit-time
- Error message tone (friendly vs. technical)

**Progress:**
- Show steps (1 of 3) vs. no indicator
- Multi-step vs. single page

## Email Tests

**Subject Line:**
- Curiosity vs. benefit vs. urgency
- Emoji vs. no emoji
- Short (<40 chars) vs. long

**Sender:**
- Company name vs. person name
- Founder vs. marketing team

**CTA:**
- Single CTA vs. multiple
- Button vs. text link
- Placement: top vs. bottom

## Test Prioritization (PIE Framework)

Score each test idea:
- **Potential:** How much improvement could this make? (1-10)
- **Importance:** How valuable is this page/element? (1-10)
- **Ease:** How easy to implement? (1-10)

**PIE Score = (Potential + Importance + Ease) / 3**

Sort by PIE score, test highest first.

## Sample Size Calculator

Don't run tests without calculating required sample size.

**Formula:**
```
n = (Z² × p × (1-p)) / E²

Where:
Z = 1.96 (for 95% confidence)
p = baseline conversion rate
E = minimum detectable effect (e.g., 0.05 for 5% relative improvement)
```

**Use a calculator:** [Optimizely Sample Size Calculator](https://www.optimizely.com/sample-size-calculator/)

## Common Mistakes

1. **Peeking at results early** (increases false positives)
2. **Not running long enough** (need statistical significance)
3. **Changing test mid-flight** (invalidates results)
4. **Testing too many things** (can't tell what worked)
5. **Ignoring guardrail metrics** (improve conversion but hurt quality)

## Further Reading

- *Don't Make Me Think* by Steve Krug (UX)
- *Influence* by Robert Cialdini (psychology)
- *How to Lie with Statistics* by Darrell Huff (avoid bad data)
- ConversionXL blog (case studies)