Skip to main content
← All case studies
Case study · Supply chain

Multi-tier supplier risk before disruption hits the P&L

A global industrial manufacturer unified supplier telemetry, lead-time forecasts, and alert routing across 140+ nodes.

Industry
Industrial manufacturing
Profile
Global OEM · $2.4B revenue
Timeline
Live in 52 days · Snowflake + SAP
85%
forecast accuracy
11d
earlier warnings
$3.1M
inventory savings

Summary

Operations teams were discovering supplier stress a week too late — after expedite fees and line-down events were already booked. Predicta ranked supplier failure risk daily and pushed executive-ready narratives to planners.

The challenge

Forecasting lived in forty regional spreadsheets. Inbound OTIF, customs dwell, and tier-2 capacity signals never met in one model. When a Southeast Asia packaging vendor slipped, planners learned from customer chargebacks — not from upstream telemetry.

What we deployed

  1. 01Ingested SAP IBP purchase orders, carrier EDI events, and Snowflake feature store history without replacing the ERP workflow.
  2. 02Deployed tier-level anomaly models tuned for volatile lead times — not smoothed monthly averages.
  3. 03Routed ranked risk signals to category managers with scenario libraries (single-source loss, port congestion, FX shock).
  4. 04Published weekly executive briefs with confidence bands instead of dashboard sprawl.
SnowflakeSAP IBPPredicta streaming alertsSlack + PagerDuty

Outcomes

  • Forecast accuracy on critical SKUs improved from 68% to 85% within one planning cycle.
  • Average disruption warning moved 11 days earlier versus the prior manual process.
  • Inventory carrying cost down $3.1M annualized in the first two quarters after go-live.
Discuss a similar deployment →