# Experiment decision memo EDM-HOLD-20260823

**Status:** Illustrative first-party memo, not a customer or production experiment  
**Run date:** 2026-08-23  
**Decision:** HOLD  
**Owner:** Illustrative design owner; no real launch authority implied

## Recorded outputs

- Assignment: 86,000 users; 50.1% control / 49.9% treatment.
- Primary: checkout completion +1.8 percentage points.
- Guardrail: refund rate +0.4 percentage points.
- CUPED: prior 28-day orders named, but rows and calculations unavailable.
- Peeking: no unofficial looks recorded; look log unavailable.

## Decision rationale

Hold because the adverse refund guardrail is visible and cannot be dismissed by the primary movement. The figures are illustrative and not inferential results.

## Attachments and limitations

The [summary CSV](https://infinisynapse.com/blog-media/experiment-decision-memo/downloads/decision-summary-EDM-HOLD-20260823.csv) exposes five aggregates and two held items. The [checker](https://infinisynapse.com/blog-media/experiment-decision-memo/downloads/verify-EDM-HOLD-20260823.py) validates displayed values only.

Assignment/outcome rows, metric definitions, SQL, interval bounds, p-values, variances, CUPED calculations, look log, and signature evidence are unavailable. This memo is not an independent audit, randomized trial, benchmark, customer result, or proof of lift.