Ab testing
A/B Testing Fundamentals
A/B testing is a method of comparing two versions of a webpage or element to determine which performs better. This guide covers the fundamental concepts you need to understand before running your first test.
What is A/B Testing?
A/B testing, also known as split testing, is a controlled experiment where you:
- Create two versions of a page or element (A and B)
- Split your traffic randomly between the two versions
- Measure performance using predefined metrics
- Analyze results to determine which version performs better
Why A/B Test?
Remove Guesswork
Instead of making decisions based on opinions or assumptions, A/B testing provides data-driven insights into what actually works with your audience.
Increase Conversions
Even small improvements in conversion rates can have significant impacts on your business. A 1% improvement could mean thousands of additional customers.
Reduce Risk
Testing changes on a portion of your traffic before rolling them out to everyone minimizes the risk of negative impacts.
Learn About Your Audience
A/B tests reveal preferences and behaviors of your specific audience, not just general best practices.
Key Components of an A/B Test
Hypothesis
Every test should start with a clear hypothesis:
"If I change [variable] to [variation], then [metric] will [increase/decrease] because [reasoning]."
Example: "If I change the CTA button text from 'Learn More' to 'Start Free Trial', then click-through rate will increase because it creates clearer value and urgency."
Variables
The element you're testing. Common variables include:
- Headlines and copy
- Call-to-action buttons
- Images and videos
- Page layouts
- Forms
- Colors and fonts
Metrics
How you measure success. Metrics fall into two categories:
Primary Metrics (what you're optimizing):
- Conversion rate
- Click-through rate
- Sign-up rate
- Revenue per visitor
Secondary Metrics (additional insights):
- Bounce rate
- Time on page
- Pages per session
Sample Size
The number of visitors needed to detect a meaningful difference. This depends on:
- Current conversion rate
- Minimum detectable effect
- Statistical significance level (95% is standard)
- Statistical power (80% is typical)
Types of A/B Tests
Simple A/B Test
Compares two versions of a single element.
Example: Testing two different headlines on a landing page.
Pros:
- Easy to set up and analyze
- Clear, actionable results
- Fast to implement
Cons:
- Tests only one element at a time
- May miss interaction effects
Multi-Element Test
Test multiple elements on the same page, where each element is tested independently.
Example: Testing headline variations AND button variations on the same page as separate tests.
Pros:
- Faster than running tests sequentially
- Independent analysis for each element
- Maximizes testing velocity
Cons:
- Doesn't capture interaction effects between elements
- Requires more traffic than single-element tests
Setting Up Your Test
1. Define Your Goal
What specific business outcome are you trying to improve?
- More sign-ups?
- Higher purchase rate?
- Increased engagement?
- Better click-through?
2. Develop Your Hypothesis
Based on user research, analytics data, or conversion optimization best practices.
Good hypothesis: "Changing the CTA from 'Submit' to 'Get Started Free' will increase conversions because it emphasizes value and removes friction."
Bad hypothesis: "Let's try a different button color and see what happens."
3. Choose Your Variable
Start with elements likely to have the biggest impact:
- Headlines (often highest impact)
- Call-to-action buttons
- Value propositions
- Hero images
4. Create Your Variations
Make meaningful differences that your audience will notice:
Good:
- "Start Your Free Trial" vs "Get Started Free Today"
- Blue button vs High-contrast orange button
- Benefit-focused headline vs Feature-focused headline
Not recommended:
- "Start Your Free Trial" vs "Begin Your Free Trial" (too similar)
- Minor color shade changes (not noticeable)
- Small copy tweaks (not impactful)
5. Set Success Metrics
Choose metrics that:
- Align with your business goals
- Can be measured reliably
- Reflect real user value
6. Launch and Monitor
Let Keak handle the statistical analysis. The test will run until:
- Significance is reached (winner found)
- Futility is detected (no meaningful difference)
- You manually stop it
Common Mistakes to Avoid
Testing Too Many Things
Problem: Testing headline, image, AND button all at once makes it impossible to know what drove results.
Solution: Test one primary change per test for clear insights.
Stopping Tests Too Early
Problem: Declaring a winner after 100 visitors because one variant is ahead.
Solution: Wait for Keak to declare statistical significance based on SPRT.
Ignoring External Factors
Problem: Running a test during Black Friday and assuming results will hold year-round.
Solution: Consider seasonality, campaigns, and other factors. Run tests during typical periods.
Not Having a Clear Hypothesis
Problem: Random testing without reasoning rarely leads to insights.
Solution: Always start with "I believe X will improve Y because Z."
Testing Insignificant Elements
Problem: Testing elements that don't impact user behavior or conversions.
Solution: Focus on high-impact areas: headlines, CTAs, value props, hero sections.
How Long Should Tests Run?
Test duration depends on your traffic and conversion rate:
Traffic Volume
- High traffic sites: Days to weeks
- Medium traffic sites: Weeks to months
- Low traffic sites: Months to quarters
Effect Size
- Large changes: Detect faster (bold redesigns, major copy changes)
- Small changes: Need more time (minor tweaks, subtle differences)
Baseline Conversion Rate
- High conversion rates (>10%): Detect changes faster
- Low conversion rates (<2%): Need significantly more visitors
Best Practices
Start Simple
Begin with high-impact, easy-to-implement tests before moving to complex experiments.
First tests:
- Homepage headline
- Primary CTA button
- Hero image
- Value proposition
Later tests:
- Navigation structure
- Page layout
- Multi-step forms
- Checkout flow
Test Continuously
A/B testing should be an ongoing process, not a one-time activity.
Build a testing culture:
- Always have tests running
- Test new pages and features
- Revisit old tests with new insights
- Build on previous learnings
Document Everything
Keep detailed records:
- Test hypothesis and reasoning
- Launch and end dates
- Results and significance levels
- Implementation decisions
- Learnings and insights
Share Results
Communicate findings across your organization:
- Build a testing knowledge base
- Share wins AND losses
- Educate stakeholders on methodology
- Create a culture of experimentation
Learn from Failures
"Failed" tests (where neither variant wins) often provide valuable insights:
- Your audience might care about different things than you thought
- The change might not be impactful enough
- External factors might be interfering
- You might need a more radical variation
When to Trust Your Results
Trust results when:
✅ Keak declares statistical significance
✅ Minimum sample size reached (1000+ visitors)
✅ Test ran for complete business cycle (at least 1-2 weeks)
✅ Both weekdays and weekends included
✅ No major external events occurred
When to Be Cautious
Be skeptical when:
❌ Very low sample size (<500 visitors total)
❌ Test ran less than one week
❌ Major marketing campaign occurred during test
❌ Seasonal event (holiday, sale) affected traffic
❌ Technical issues occurred (site downtime, tracking errors)
Practical Significance vs Statistical Significance
A result can be statistically significant but not worth implementing.
Example:
- Variation B wins with 99% confidence
- Improvement is 0.1% conversion rate increase
- Implementation requires major development work
- Projected revenue gain: $200/year
- Development cost: $5,000
Decision: Statistically significant, but not practically worth it.
Always consider:
- Business impact vs implementation cost
- Long-term sustainability
- Brand alignment
- User experience implications
Next Steps
Now that you understand A/B testing fundamentals:
- Launch Your First Test - Get started with testing
- Learn About SPRT - Understand how Keak determines winners
- Understand Test Types - Choose the right test for your goals
Ready to start testing? Open the Keak extension on your website and create your first variation.