From a Snapshot to a Video
General Mills, and the Model That Taught Before It Ran
General Mills had capable engineers analysing scenarios with figures and spreadsheets, and the answers kept falling short. What changed was not the speed of the arithmetic. It was that a dynamic model let them watch the system behave — and the most valuable part arrived before the first result did.
It started as a manufacturing question
The engagement most people remember is the supply chain one, but it began smaller and squarely on the plant floor. Five years earlier, General Mills used the same simulation approach at a single site to settle a capital decision: how many packaging lines to buy for a particular product line.
That is the ordinary shape of a manufacturing capital question, and it is the shape ReliaSim is built for. Buy too few lines and you throttle a product. Buy too many and you have stranded capital that shows up as utilisation nobody can defend. The answer depends on variability and interaction, not on average rates.
Why the spreadsheet fell short
Later, the task was extracting cost from a major product line by refining the production system, and it needed something the old method could not give. The engineers had been analysing scenarios at an aggregate level with figures and spreadsheets. As Mike Geddis, director of engineering, put it, the company needed a tool that could model complex relationships across the whole system and draw “the feasible kind of conclusions that helped whittle down alternatives.”
“Instead of taking a snapshot, this is more like a video. It’s more than just a point solution. Your ability to actually see the activity in action and plot out the results allows you to understand impacts much better while the supply chain is running. We can run six months of supply chain data in a matter of minutes.”— Mike Geddis, Director of Engineering, General Mills
The finding nobody plans for
The impact arrived earlier than expected — not when the model produced results, but while it was being built.
“What we really found to be very powerful was the actual building of the model. It forced a certain diligence in terms of saying, if we change the system here this happens over here. We had not gone to that level of understanding the interconnections of our systems before.”— Mike Geddis, Director of Engineering, General Mills
This is worth sitting with, because it is not a software feature. Building a model that runs forces every hand-waved connection in a plant to be made explicit and consistent. Engineers who had worked in that system for years found they had never been required to state how one change propagated to another. The model made them state it.
What it was worth once running
Outcome
| Measure | Value |
|---|---|
| Six months of supply chain data | minutes |
| Time to a result, once the model existed | as little as 15 minutes |
| Alternatives explored | substantially more than before |
“This definitely saves us time. It gives the ability to let the tools follow our questions and follow them rapidly, so once we have the model constructed, we can get results in as little as 15 minutes. There’s no question we will explore substantially more alternatives than we would have if this were a more laborious process like before. As a result, we end up with much more thorough and novel solutions.”— Mike Geddis, Director of Engineering, General Mills
The phrase to notice is “let the tools follow our questions.” Speed is not interesting because it saves an afternoon. It is interesting because below a certain answer time, people stop rationing their questions — and the quality of the final decision depends far more on how many alternatives were genuinely tested than on how precisely any one of them was computed.
The method
The work was done by the same team using Supply Chain Builder. This is a supply-chain engagement with a manufacturing core: it began as a plant capital decision and grew outward to multiple plants, product portfolios and locations. ReliaSim addresses the production-system half of that picture; the network half is SCModeling.
Published as IIE Solutions, June 2001 (PDF), reprinted with permission of the Institute of Industrial Engineers.
Related: Spreadsheet, AI, or ReliaSim? · The Hidden Bottleneck · All case studies