Experimentation Leader.
Full-Stack Growth Marketer.

Atticus Li — and the experiments he's run

Every article is published under my byline because the experiment data, the behavioral economics interpretation, the methodology calls, and the editorial line are mine. I'm a CXL-certified CRO practitioner, one of ~1,000 Mindworx-certified behavioral economists worldwide, and I've led 200+ A/B tests at NRG Energy, Silicon Valley Bank, and across SaaS, fintech, e-commerce, and marketplace verticals. I don't ghost-write, don't outsource to anonymous contractors, and don't publish anything I wouldn't defend in a stakeholder review.

Atticus's research, frameworks, and interpretation — with AI used as a drafting tool

The substance of every article — the experiments, the behavioral economics interpretation, the frameworks, the methodology calls — is Atticus's. AI assistance, where used, is a tool for research synthesis and prose drafting, not a content generator. Articles are produced through a multi-step pipeline he built and operates:

1. Topic selection — I pick the topic from one of three sources: a running A/B test in my portfolio, a behavioral economics principle I want to interpret for practitioners, or a question that keeps coming up in mentoring calls.
2. Research — Claude (Anthropic's LLM) drafts a research brief: relevant academic studies, current statistics, industry positions. I review and correct it.
3. Experiment data — I supply anonymized results from my own portfolio in lift ranges, never exact percentages, never identifiable brands or page names. The behavioral interpretation of why the result happened is mine.
4. Drafting — Claude produces a first draft in my voice using a system prompt that encodes my writing style, mental models, anti-patterns, and data-protection rules.
5. Human review — I read every article before publishing. I rewrite sections that don't match how I'd actually frame it, kill claims I can't defend, fix anything that smells generic, and add specifics from my own experience that the AI couldn't have known.
6. Publish — Only after human review does an article move from draft to published in Sanity CMS.

Why disclose this at all? Because being upfront about the writing process is the right thing to do, and because it's the difference between "AI generated this for me" (which would be misleading — the substance is mine) and "AI helped me write faster about my own research" (which is the truth). The disclosure lives here so the byline can stay focused on what actually matters: the experiments, the interpretation, and the frameworks the article is built from.

To help practitioners, not to game search rankings

This site exists because there's a gap between the academic behavioral economics literature and the day-to-day reality of running an experimentation program. I write the interpretation that connects Kahneman, Thaler, and Cialdini to the test you have to ship next quarter under a budget constraint with a stakeholder who doesn't trust statistics. If an article isn't useful to a working CRO manager, product manager, or growth lead — I don't publish it.

What every article has to clear before it ships

• Non-commodity insight — the article has to add something a generic "7 tips" piece couldn't. Usually that's a behavioral economics interpretation, a methodology distinction, or a pattern observed across many tests.
• First-hand specifics — at least one anonymized data point, lift range, or experiment observation from my own work. No second-hand case study summaries.
• Data protection — no exact percentages, no internal test IDs, no brand names, no specific page descriptions that could identify a client. Ranges and generic descriptors only.
• Intellectual honesty — if a result was inconclusive, I say so. If a behavioral principle didn't transfer to the context, I say that too. The replication crisis is real; faking certainty erodes trust.
• Practitioner usefulness — a reader should be able to apply something concrete within a week. If the article only flatters my expertise, it gets cut.

How to flag an error

If you spot a factual mistake, a misattributed study, a broken claim, or any place where the writing doesn't hold up — email me at atticus@atticusli.com with the URL and the specific line. I'll review and correct, then add a dated correction note to the article. I take this seriously; if my byline is on something, I own it.