Hi, I'm Tilman.
I spend most of my time figuring out why users do things we didn't expect.
Worked with, among others
What I've learned
The best results start with a problem.
ALB Filter | eCommerce
+8.5% revenue per user
helpcheck | Landing page
+23% sign-ups
123fahrschule | Navigation
14% more accurate, 30% more direct, 37% faster
LET'S DO THE MATH
Is it even worth it?
Sometimes user research beats running another test. Here you can estimate what's more realistic for your shop.
Your numbers
The smallest improvement you want to be able to detect. 5% is standard.
A/B testing works, with patience
You can run tests, but expect longer runtimes. Focus on ideas with big impact.
Rough revenue opportunity
+€86,400/year
Tests per year
5
~1 expected winners
Duration per test
~9 weeks
until significance
For stats nerds
Required sample size411,600 visitors (205,800 per variant)
Your monthly revenue€320,000
Min. detectable effect5%
Confidence level90%
Assumed win rate20%
False-positive discount10%
The yearly uplift assumes every winning test generates 5% additional revenue for ~6 months on average, discounted for false positives at a 90% confidence level.