Module 03 · Match Intelligence · 03C

Match Rules Studio

How do thresholds and tolerances change what we accept?

Try this
  1. 1Tighten beauty size tolerance from ±8% to ±4%
  2. 2Run test-on-sample
  3. 3Compare before/after distribution, then Apply or Roll back
  4. Expected: 236ml vs 250ml (5.9% gap) falls out of band; auto-accept share drops before you apply.
Draft configuration

Thresholds & weights

Auto-accept vs review distribution

Current applied rules

Auto-accept rate
48.4%
Review
48.6%
Flag
3.0%
Beauty size band test

236ml vs 250ml mist

Size gap 5.9% vs tolerance ±8% → in band, score 0.81
Candle band: 360–411g (14.2% spread) needs ±15% to auto-pair large candles.
Arabic normalisation

Text pipeline examples

أ / إ / آ → اإكسسواراتاكسسوارات
ة → هشمعة كبيرةشمعه كبيره
ى → يمستوىمستوي
TransliterationOud ↔ عودoud | عود
Seeds

Category-to-category mapping, fit & food component rules

Beauty
mist↔mist · lotion↔lotion · large candle 360–411g · tumbler 170–198g · soap↔hand wash
Apparel
gender must match · fit exact under strict rule · fabric −0.12 · wash −0.08
Food
component-level · portion must match · LTO/combos excluded from core SPI
PricePulse · Competitive Pricing Intelligence OS · Simulated demo • Fictional data. All brands, competitors, people, prices and dates are invented.Powered by Quantzig