Skip to main content
A good control recipe starts with one question and finishes with evidence you can explain. Use these recipes after the first-client journey, or jump straight to the expectation you need.

Read a recipe in four parts

  1. Scope: which clients and subjects are assessed?
  2. Condition: which observation is compared with which expectation?
  3. Examples: what would satisfy, contradict or leave the question unanswered?
  4. Limits: what requires another control or a manual review?
The simple control editor can create an observation-based population and a single expectation. Exact population values and additional filters require an appropriate built-in definition or an advanced definition.

Before enabling a recipe

Confirm source coverage, inspect an editable draft, review client settings and test the result cases. Sample thresholds are starting points to discuss, not a recommended policy for every client.