2/ 10Early state
A: 9 leads from 80 visits, B: 12 leads from 80 visits. Not significant.
Then
The test is complete. You read the result in Peak Experiments.
Keep it running · no winner
Peak Pages · A/B testing · Demo data
Two variants. One reliable answer.
Instead of a gut feeling.
Test page variants against each other. Each visitor always sees the same variant. Significance and the decision run through Peak Experiments, and you promote the winner yourself.
Prepare an A/B test in Peak PagesDemo data · fictional numbers · no real customer results
DEMO DATA / A/B TESTOne example value.
One example value.
The fitting next step.
01 / Test state
How far along is the test? (runtime 0 to 10 of 10)
02 / What it meansA: 9 leads from 80 visits, B: 12 leads from 80 visits. Not significant.
Keep it running · no winner
Switch between an early and a late test state. All numbers are demo data.
Demo data · no real test
One example. Two paths.
One setting
changes the page.Compare both paths below ↓
changes the page.Compare both paths below ↓
Two test states · demo data
How far along is the test? (runtime 0 to 10 of 10)
Demo data: invented numbers. In the product Peak Experiments calculates significance, not this page.
10/ 10Late state
The test is complete. You read the result in Peak Experiments.
You decide · publish the better variant yourself
Trust needs clarity
Honest about tests.
01 / The existing evidence
Demo data
The numbers above are invented and show no real customer and no real winner.
- 2/10A: 9 leads from 80 visits, B: 12 leads from 80 visits. Not significant.Keep it running · no winner
- 10/10The test is complete. You read the result in Peak Experiments.You decide · publish the better variant yourself
02 / Your check
No winner without significance
Peak Pages does not declare a winner just because one variant is ahead. The result comes from Peak Experiments.
4 / 1010 / 10
Compare both paths and their reasons.
03 / The clear limit
You decide
Peak does not apply even a significant result on its own. You publish the winner.
Демо│Nothing published
The next step
Improve what demonstrably works.
Start Peak Pages, add a second variant and let Peak Experiments check the result.
Prepare an A/B test in Peak PagesPeak Pages · A/B testing · Demo dataOne change.
Demo data · fictional numbers · no real customer results
Frequently asked questions
Does a visitor always see the same variant?
Yes. The server assigns each visitor to one variant for good.
Who decides on the winner?
You do. Peak Experiments calculates significance; nothing is applied on its own.
Are the test numbers here real?
No, they are demo data with no real customer and no real winner.