Investigate the next budget move

Business result

Portfolio contribution within the agreed cash-exposure limit.

Next step

Design a controlled test

Working output

Project brief

Unit of analysis

Channel × market × proposed budget step × completed cohort

Decision owner

Head of User Acquisition

Controlled test design

• Hypothesis: Test one bounded budget change with stop and restore conditions.
• Comparison: Concurrent unchanged baseline versus budget step; track spillovers and business outcomes
• Specify assignment, group overlap, exposure and contamination checks.
• Choose a primary metric, detectable effect and completed outcome window before calculating sample size.
• Set risk budget, guardrails, checkpoint and stop or restore conditions.

Acceptance checks

• Historical decisions use no future information.
• Search costs, error costs and uncertainty are included; counterfactual upside is estimated separately.
• Decision owner, risk budget, guardrails, checkpoint and restore condition are specified.
• Assignment, detectable effect and outcome horizon are specified before the test starts.

Deliverables

• A pilot design with success and stop criteria
• Agreed scope, source requirements, responsibilities and acceptance checks.
• A working brief for a readiness and delivery conversation, not a completed investigation.