experiment-designer

To design an A/B test for Adobe Edge Delivery Services, you convert business hypotheses into testable variant setups using either page metadata or an experiments spreadsheet, defining success metrics and traffic splits. This produces a structured experiment brief for your EDS pages.

Calculating sample size and duration for statistical significance involves defining your primary metric and Minimum Detectable Effect (MDE). The process generates an experiment brief with specific duration targets and a monitoring plan to prevent premature or biased conclusions.

Yes, you can use an experiments spreadsheet like `/experiments.json` instead of page metadata for EDS experimentation. The configuration guidance includes validation rules for variant URLs, traffic splits, and dates to ensure your setup is correct.

Setting up a multivariate test with a specific traffic split requires defining variant URLs and validating the distribution, such as a 50/25/25 split. The configuration checks variant parity and validates metadata to ensure correct traffic allocation.

Limitations for CDN edge experiments include enforcement of safety constraints against unsafe fetching and unsupported workflows. The configuration process validates metadata and variant parity to ensure only supported EDS page-level experimentation scenarios are executed.

The best way to structure an experiment brief for multivariate testing is to include the setup, decision criteria, and common pitfalls. It should define success metrics, MDE, control and variant intent, and a monitoring plan to ensure statistically valid results.