ab-variant-creator

name	ab-variant-creator
description	Use when creating A/B test variants for headlines, copy, emails, or landing pages. Helps generate meaningfully different test variants with clear hypotheses and expected outcomes.

A/B Variant Creator

Overview

Create A/B test variants that actually teach you something. This skill helps you design tests with meaningful differences, clear hypotheses, and statistically significant potential—not just random variations.

When to Use

• Testing headline variations
• Comparing email subject lines
• Optimizing landing page copy
• Testing different value propositions
• Comparing CTA approaches
• Validating messaging angles

Output Structure

TEST OBJECTIVE: [What you're trying to learn]

HYPOTHESIS: "If we [change], then [metric] will [improve/decrease] because [reason]."

---

CONTROL (A):
[Current or baseline version]

VARIANT B:
[First test variant]
- Change type: [Headline/CTA/Value prop/Tone/Length]
- Hypothesis: [Why this might win]

VARIANT C (optional):
[Second test variant]
- Change type: [Different element or approach]
- Hypothesis: [Why this might win]

---

TEST PARAMETERS:
- Primary metric: [CTR/Conversion rate/etc.]
- Secondary metrics: [Other things to watch]
- Minimum sample size: [Calculated estimate]
- Test duration: [Recommended timeframe]
- Confidence level: [95% standard]

EXPECTED OUTCOMES:
- If A wins: [What it tells you]
- If B wins: [What it tells you]
- If C wins: [What it tells you]
- If inconclusive: [Next steps]

Quick Reference

Element	Best Practice
Variable count	Test ONE variable per test
Difference level	Meaningful, not minor tweaks
Sample size	1,000+ per variant minimum
Duration	1-2 full business cycles
Hypothesis	Write before testing
Winner action	Implement, then iterate

Test Types

Headline Tests

• Benefit vs. feature focus
• Question vs. statement
• Short vs. long
• Specific vs. general

CTA Tests

• Action words (Get vs. Start vs. Claim)
• First person vs. second person
• With urgency vs. without
• Length variations

Value Proposition Tests

• Lead benefit vs. lead pain point
• Emotional vs. rational appeal
• Social proof presence vs. absence
• Risk reversal vs. no mention

Format Tests

• Long copy vs. short copy
• Bullets vs. paragraphs
• Video vs. text
• Form length variations

Common Mistakes

• Too small changes: "Get Started" vs "Get Started Now" won't teach much
• Multiple variables: Changing headline AND CTA AND image = no learnings
• Stopping too early: Wait for statistical significance
• No hypothesis: Random testing without learning goals
• Ignoring segments: Overall winner may lose in key segments
• Testing opinions: Test strategic questions, not personal preferences

Examples

Example 1: Landing Page Headline Test

Input: Create A/B test variants for a productivity app landing page. Current headline: "The Best Task Management App"

Output:

TEST OBJECTIVE: Determine if benefit-specific headlines outperform generic claims

HYPOTHESIS: "If we replace generic superlative ('Best') with specific outcome ('Finish work 2 hours earlier'), then signup rate will increase by 15-25% because specificity builds credibility and relevance."

---

CONTROL (A):
"The Best Task Management App"

VARIANT B:
"Finish Work 2 Hours Earlier Every Day"
- Change type: Generic claim → Specific outcome
- Hypothesis: Concrete benefit is more believable and desirable than subjective "best"

VARIANT C:
"Stop Forgetting Tasks. Start Finishing Projects."
- Change type: Generic claim → Pain point + outcome
- Hypothesis: Addressing the pain directly creates emotional connection and relevance

---

TEST PARAMETERS:
- Primary metric: Free trial signup rate
- Secondary metrics: Time on page, scroll depth, bounce rate
- Minimum sample size: 2,000 visitors per variant (6,000 total)
- Test duration: 2 weeks or until significance reached
- Confidence level: 95%

EXPECTED OUTCOMES:
- If A wins: Audience values brand positioning over specifics (rare, retest)
- If B wins: Specific outcomes resonate—test more specific claims
- If C wins: Pain-point messaging works—explore problem-aware copy
- If inconclusive: Test bigger differences (format, social proof)

Example 2: Email Subject Line Test

Input: Create A/B test variants for a SaaS onboarding email. Current subject: "Welcome to TaskFlow"

Output:

TEST OBJECTIVE: Determine if action-oriented subjects increase open rates over welcome messages

HYPOTHESIS: "If we replace passive welcome with action prompt, then open rate will increase by 20-30% because it creates curiosity and implies immediate value."

---

CONTROL (A):
"Welcome to TaskFlow"

VARIANT B:
"Your first project is waiting (start here)"
- Change type: Welcome statement → Action prompt
- Hypothesis: Implying something is ready for them creates urgency and curiosity

VARIANT C:
"TaskFlow setup: 3 steps, 5 minutes"
- Change type: Welcome statement → Specific time/effort commitment
- Hypothesis: Showing low effort reduces procrastination on opening

---

TEST PARAMETERS:
- Primary metric: Email open rate
- Secondary metrics: Click rate, time to first project creation
- Minimum sample size: 500 recipients per variant (1,500 total)
- Test duration: 1 week (or 1,500 new signups)
- Confidence level: 95%

EXPECTED OUTCOMES:
- If A wins: Brand warmth matters more than action prompts in onboarding
- If B wins: Curiosity/implied value drives opens—test more teasers
- If C wins: Effort clarity reduces friction—be specific in future emails
- If inconclusive: Subjects aren't the bottleneck—test send timing or preview text

Example 3: Pricing Page CTA Test

Input: Create A/B test variants for a pricing page CTA. Current CTA: "Start Free Trial"

Output:

TEST OBJECTIVE: Determine if addressing objections in CTA increases trial starts

HYPOTHESIS: "If we add 'No Credit Card' to the CTA, then conversion rate will increase by 10-20% because it removes the primary signup hesitation at point of action."

---

CONTROL (A):
"Start Free Trial"

VARIANT B:
"Start Free Trial—No Card Required"
- Change type: CTA only → CTA + objection removal
- Hypothesis: Addressing credit card concern at decision point reduces friction

VARIANT C:
"Try Free for 14 Days"
- Change type: Action focus → Time-specific framing
- Hypothesis: Specific trial length sets clear expectation and feels lower risk

---

TEST PARAMETERS:
- Primary metric: Trial signup rate
- Secondary metrics: Trial to paid conversion (watch for quality)
- Minimum sample size: 1,500 pricing page visitors per variant
- Test duration: 3 weeks or until significance
- Confidence level: 95%

EXPECTED OUTCOMES:
- If A wins: Simplicity wins—avoid cluttering CTAs
- If B wins: Objection handling in CTA works—test other objections
- If C wins: Time framing matters—test different durations (7, 14, 30)
- If inconclusive: CTA isn't bottleneck—test page layout, testimonials, pricing display

Quality Checklist

• Clear hypothesis written before variants
• Only ONE variable changed per test
• Variants are meaningfully different (not minor tweaks)
• Sample size calculated and realistic
• Primary metric clearly defined
• Expected learnings documented for each outcome
• Test duration accounts for business cycles
• Secondary metrics identified
• Next steps planned for each outcome
• Inconclusive scenario has follow-up plan