Atticus Li — Better Decisions

In 1962, a former Stanford MBA student named Phil Knight wrote a master's thesis arguing that Japanese running shoes could disrupt the American athletic-footwear market the same way Japanese cameras had disrupted Leica.

If you've ever booked a flight on Kayak or Expedia and watched the "Searching 137 airlines..." progress bar churn for three seconds before showing you results, congratulations — you've personally experienced one of the most carefully-engineered behavioral…

In 2017, a Netflix engineering team published a quiet little blog post on the Netflix Tech Blog titled "Artwork Personalization at Netflix."

In the late 1940s, Chivas Regal was dying.

In the summer of 1997, Apple was somewhere between 60 and 90 days from bankruptcy.

In the mid-1990s, a British perfumer named Jo Malone got Bergdorf Goodman in New York to host a pop-up shop for her brand.

Here is a trick almost no marketer learns properly in business school: in many product categories, higher prices increase demand instead of decreasing it.

Here’s a fact that should reshape how you think about every product, restaurant, vacation, and customer experience in your life: The single most important paper in modern behavioral economics is about colonoscopies.

If you’ve ever watched Mad Men and thought surely the ad guys can’t really be that powerful — it’s just a show, I’m here to gently disagree.

There’s a particular blog post format you’ve read a dozen times: “17 cognitive biases ruining your conversion rate.”

In the autumn of 1927, a young Russian psychology student named Bluma Zeigarnik went to lunch at a busy restaurant in Berlin.

Something interesting happened quietly in a recent ad platform API release: experiment statistics got pulled directly into the same reporting layer as campaign performance data.

Most teams treat the moment a test hits significance like a gun going off at the end of a race.

A hypothesis is not a prediction.

There's a moment in most experimentation programs when volume becomes the goal.

Outcome-dependent stopping is a common way to invalidate a fixed-horizon test.

When a result is selected because it is extreme, a repeat measurement often lands closer to the underlying mean.

New CRO analysts learn the field largely through published case studies.

These five statistical failure modes recur in new CRO programs.

A bare reported lift is uninterpretable.

Steve Krug published the third edition of _Don't Make Me Think_ in 2014.

In 2019, a research team at Princeton's Center for Information Technology Policy did something nobody had done at scale before.

In 2011, Netflix shipped a flow that ran counter to every retention playbook in the streaming industry.

I've audited a lot of pricing pages.

A vague benefit badge creates more friction than no badge at all.

Time-on-page is not directional on its own.

Mobile and desktop are different products.

This article began as an analysis of a historical working export with more than 200 rows.

A CTA's click rate measures visibility.

A new CTA producing a small positive topline estimate is not automatically a real win.

A CTA's click rate is not its conversion contribution.

Adding another CTA to an already crowded hero creates a testable risk: it may redistribute attention rather than create new demand.

Most teams discover A/B testing and immediately want to run more experiments.

If you only read one paragraph of this, read this one: "Fail to reject" does not mean "there is no difference."

There's a pattern worth noticing every time a new category of "automated optimization" software launches: the marketing promises to replace the hard thinking, and teams rush to buy the tool before they've diagnosed why their current testing isn't working.

Every few months, a new platform promises to automate conversion optimization.

Three years into running a formal experimentation program, I made a change that felt almost heretical: I stopped treating every experiment like a coin flip that needed to run to 95% statistical significance.

Resurfacing Old A/B Tests: A System for Faster Iteration Cycles and the Revival Value Formula TL;DR: Most old A/B tests contain insights your team no longer remembers.

How to Prevent Institutional Knowledge Loss in Your A/B Testing Program: The Tribal Dependency Index TL;DR: The worst habit that kills institutional memory isn't forgetting to document.

Turning 100 Experiments Into Strategic Insight: The Insight Density Curve TL;DR: The value of the 50th experiment isn't the same as the value of the 5th.

How to Run Meta-Analysis Across Historical A/B Test Data: The Aggregation Validity Threshold TL;DR: Meta-analysis isn't about combining experiments — it's about knowing when you have enough similar tests for the aggregate to tell you something true.

Experimentation Management Systems: The Process Maturity Index for Mature Teams TL;DR: Running a lot of A/B tests isn't maturity.

Building an Experimentation Knowledge Base for High-Velocity Growth Teams: The Knowledge Compounding Rate TL;DR: A knowledge base doesn't just store past experiments — it's how data beats the HiPPO in decision-making.

Centralized A/B Testing Database Design: The Trace Ratio That Predicts Program Health TL;DR: A centralized A/B testing database is only as useful as the fraction of experiments you can fully reconstruct from it.

Best A/B Test Library Software: The Tool Fit Matrix for Evaluation TL;DR: The "best" A/B testing platform isn't a single tool — it's the one that fits your team's scale, statistical needs, integration stack, and cost curve.

How to Avoid Repeating Failed Experiments: The Failure Recurrence Rate TL;DR: Repeated failed experiments aren't a sign of ambition — they're a sign the team isn't reading its own archive.

A/B Testing Documentation Framework: Templates, Metadata Standards, and the Hypothesis Reuse Rate TL;DR: The point of documenting experiments isn't to record what happened.

A/B Test Repository Architecture: Schema, Tagging, and the Retrieval Time Budget TL;DR: A/B test repositories don't fail because the schema is wrong.

Organizing Experiments Across Product, Marketing, and UX: Measuring the Coordination Tax TL;DR: Siloed experimentation isn't a culture problem.

CRO Knowledge Management: Building Institutional Memory Against the Knowledge Half-Life TL;DR: Every experimentation program has a Knowledge Half-Life — the speed at which insights from past tests become inaccessible.