In this case study learn how BRIDGEi2i helped a Fortune 100 Technology company to recognize anomalies in historical bookings as outliers and to treat them in statistically acceptable ways for better forecasting and demand.
Demand Planning for Big Deals (Fortune 100 Technology Company)
1. A Case Study in
Demand Planning for
Big Deals
A Fortune 100 Technology Company
Quick Context
Objective
âą 1% higher forecast
accuracy; ~$30mn
business impact
âą Better Revenue
Planning enabled
due to removal of
forecasted
anomalies
Impact
âą BRIDGEi2i has developed numerous
ways of dealing with Outliers - a key
element in forecasting
âą We bring our experience & knowledge
of best practices in other industries to
our clients
Key Success Elements
Our Approach
3 Months
3 Years
Client
Project length
Length of relationship with client
âą All data was securely accessed within
the client environment
âą SAS was used for the Anomaly
Detection algorithm development and
deployment
âą SFDC data was used to recognize
Large Deals based on business rules
âą When Large Deals happen or fall-off,
demand peaks are observed â Order
Data was used to recognize outcome
âą Anomalies happen when Large Deals
go through or fall off
âą An algorithm was used to generate a
Risk Score for each Large Deal
âą The Risk Score â a factor of Moving
Average Demand based on product &
customer attributes â was used to
deflate the Deal Size
âą High Risk deals were excluded while
forecasting demand
âą A rigorously tested code was developed
and validated repeatedly on historical
Bookings prediction accuracy
âą The final SAS code would fetch data
from SFDC, Order Data and historical
Bookings, Identify and flag outliers in
Demantra â the single platform for
demand intelligence for the Planners
âą Model has yielded great results; ~80%
adoption by Demand Planners
Data Management Algorithmic Play Operationalization
a. ~40,000 SKUs and a global dynamic demand scenario; very volatile demand
b. Short product lifecycles and highly competitive landscape
a. To recognize anomalies in historical bookings as outliers
b. To treat them accordingly â in statistically acceptable ways for better
forecasting of demand