Skip to content
Documentation
All docs

Tools

Policy Simulator

Replay historical returns against proposed thresholds before applying them.

The Policy Simulator lets you propose a configuration change — most commonly a risk threshold — and replay your last 30 days of returns against it, before you commit. It’s the safest way to tune scoring without learning the hard way that your new thresholds were too aggressive.

How it works

  1. Pick a simulation type. Today this is SCORING_THRESHOLD— the LOW / MEDIUM / HIGH zone boundaries.
  2. Enter the proposed values. The current values appear next to each input.
  3. Choose a replay period (default 30 days).
  4. Run the simulation. The server replays every scored return through the proposed thresholds and computes the new zone for each.

What you get back

  • Before / after comparison— A grid showing how many returns moved between zones (e.g. “48 returns downgraded from HIGH to MEDIUM”).
  • Affected returns table— Every individual return that changed zone, with its score, customer, and the new zone.
  • Apply or discard— If the preview looks right, click Apply to Settingsand the new thresholds go live for all future scoring. If it doesn’t, click Discard and try different numbers.

What it does NOT change

  • Existing risk scores are not retroactively rewritten when you apply new thresholds. Only future scoring uses the new boundaries.
  • Signal weights and per-signal logic stay untouched. The simulator only moves the zone boundaries.

When to use it

  • You’re seeing too many MEDIUM-zone returns in your review queue. Try raising the MEDIUM boundary.
  • Your prediction accuracy on chargebacks is too low. Try lowering the HIGH boundary.
  • You changed your return policy and want the engine’s zone definitions to follow.

Next steps