The encyclopedia · Strategy & Leadership · Strategic decision · 1997
P&G rebuilt its North American supply chain with OR plus judgement, saving $250M a year.
P&G combined integer programming, network optimization and managers' judgement to restructure sourcing and distribution, saving $250M+ a year.
Procter & Gamble
the move
Procter & Gamble markets over 300 consumer brands across 70-plus countries, so redundant production and distribution is both easy to slip into and expensive to keep.
The obvious response is to move a few plants or trim a warehouse. P&G instead ran a large re-engineering of its North American sourcing and distribution, involving about 500 staff over several months.
The method combined information technology, integer programming, network optimization and a GIS, using LINDO to solve the model, then applied managers' judgement to the answer before acting — a hybrid of model and ground truth.
The overhaul of P&G's manufacturing and distribution system is saving well over $250 million a year, mostly from lower manufacturing expenses.
why it works
- The problem spans plants, warehouses and customers, so it needs one model
- A GIS made the network's geography visible and the model's answer inspectable
- Managerial judgement caught real constraints the data could not see
- The scope was corporate, so the savings were hundreds of millions a year
what transfers
A model is only as good as the real constraints it encodes; the highest-value OR pairs optimisation with the judgement of people who know the plants, customers and politics.
what came after
The redesign cut duplication in P&G's North American supply chain, saving well over $250 million a year. The 1997 paper 'Blending OR/MS, Judgement, and GIS: Restructuring P&G's Supply Chain' was published in Interfaces and became a Franz Edelman Award finalist.
references
- LINDO Helps Redesign Supply Chain
- James R. Evans author biography (Edelman finalist; P&G supply chain)
spotted an error? The archive wants to know.