ab-test-loop

What problem does it solve?

Autonomous A/B test iteration loop for PostHog experiments. Monitors all running experiments for Bayesian significance (>95%), declares winners, rolls out the winning variant to 100% via PostHog feature flags, generates the next copy variants informed by brand voice, and launches the next experiment — without manual intervention. Supports CTA button text, hero section headlines, and any other PostHog experiment type. Also handles one-time scroll depth tracking setup (25/50/75/90%) via Webflow Pages API. Use when user says "ab-test-loop", "check my a/b tests", "run the ab loop", "check experiment results", "declare ab test winner", "launch next variant", "set up scroll tracking", or "automate ab testing".

Core Features & Use Cases

• Autonomous monitoring of all running PostHog experiments and automatic significance checks.
• Bayesian-based decisioning to identify significant winners (≥ 95% probability).
• 100% rollout of winning variant via feature flags; 0% for losers to ensure clean transitions.
• Automated variant generation aligned with brand voice to inform next test iterations.
• Sequential experiment management: generate and launch subsequent experiments with updated copy.
• Phase S one-time scroll depth tracking setup for Webflow pages to capture engagement signals.
• Runs from the client workspace (clients/{domain}/) to keep context anchored to the relevant site.

Quick Start

Invoke ab-test-loop to automatically read all running PostHog experiments and begin monitoring, winner declaration, and variant generation.