experimentation-playbook

Experiment lifecycle management is handled by an LLM-driven graph that orchestrates bets from intake through design, preregistration, execution, monitoring, archival, and strategy updates. You provide a bet reference or raw hypothesis to initialize the workflow.

Preregistration in GTM and product experiments is a predefined workflow vertex that logs your hypothesis and design before execution begins. It enforces guardrails through explicit vertex transitions, ensuring rigorous tracking and predictable progress before outcomes are written to logs.

Yes, you can track pricing and onboarding bets using a graph-driven workflow. The orchestration supports experiments across GTM, product, pricing, content, and onboarding, applying explicit vertex transitions to enforce predictable progress from intake to archival.

Experiment monitoring and logging are set up by defining roles like experiment-runner, monitor-loop, and strategy-updater within the workflow. The monitor-loop vertex tracks execution, and the strategy-updater writes validated outcomes to bets, experiments, and learnings logs.

The best way to enforce guardrails during experiment design and execution is using a graph-driven workflow with declared valid transitions per vertex. This ensures rigorous lifecycle tracking by requiring explicit vertex transitions between roles like experiment-designer, preregistration, and experiment-runner.

Yes, you need to define specific roles like bet-logger, experiment-designer, and archiver to run experiments. The workflow supports these defined roles to orchestrate end-to-end lifecycle tracking, requiring explicit vertex transitions to progress from intake to strategy updates.