Ideas on optimizing every visit
Short, practical notes on AI-native CRO, personalization, and getting more from the traffic you already pay for.
The math behind Optimeleon: how a bandit serves the right page to every visitor
Optimeleon does not run a fixed split and wait for a verdict. It remakes the traffic decision on every visit using Thompson sampling. Here is the math under the hood, in plain language.
Read more →A/B testing vs multi-armed bandits: when to use which
Not a holy war. They solve different problems. Here is a clear decision guide for when a classic A/B test wins and when a bandit does.
Read more →What actually makes a landing page convert
Forget the swipe files. A landing page converts when it matches intent, makes one clear promise, proves it, and asks for the right next step. Here is the anatomy.
Read more →When your winning variant is quietly losing: Simpson's paradox in A/B tests
A variant can beat the control overall and still lose with every audience you actually care about. Here is how Simpson's paradox slips into conversion tests, and how to stop trusting the wrong number.
Read more →Designing for personalization: UX patterns that hold up
Personalized interfaces break in ways static ones never do. Here are the UX patterns that keep a shifting page coherent, fast, and trustworthy.
Read more →How to tell if your landing page copy is any good
Good copy is not clever, it is clear and earned. Here is a practical way to judge a landing page's words before you spend traffic testing them.
Read more →Personalization that starts with the customer, not the tech
Most personalization optimizes for the company's funnel. The kind that works starts from what the visitor actually came to do. Here is the difference.
Read more →Creative experimentation: testing more than the button color
Button-color tests are why people think experimentation is dead. The teams winning are testing message, angle, and structure, fast. Here is how.
Read more →Multi-armed bandits, explained without the jargon
A multi-armed bandit is just a smarter way to spend attention while you learn. Here is the intuition, minus the equations, and where it beats a classic test.
Read more →Why most of your A/B tests never reach significance
If your tests keep ending 'inconclusive', the problem usually is not the idea. It is statistical power. Here is the math of why, and what to do about it.
Read more →False positives, false negatives: Type I and Type II errors in A/B testing
Two ways an A/B test can lie to you: calling a flat result a winner, and missing a real one. Here is what each costs and how to keep both in check.
Read more →The CRO audit: a practical walkthrough
A conversion audit does not need a 60-tab spreadsheet. Here is a focused way to find where your funnel leaks, why, and what to fix first.
Read more →How to redesign your site without tanking conversions
A fresh design can quietly cost you customers. Here is how to ship a redesign that looks better and converts better, instead of trading one for the other.
Read more →Ad-to-Variant matching, explained
Your ad promised one thing. Does your landing page deliver it? Here's how keyword-aware pages lift paid conversion without more spend.
Read more →Why classic A/B testing is quietly holding you back
Two variants and a three-week wait was state of the art in 2008. Here's what changes when generation is free and the system optimizes per visitor.
Read more →