Where we left off#
- Week 10: a constrained optimiser over a real dataset
- Week 11: what that optimiser does not price in
- both were built around one shopper and one basket
- this week: fifty people, and what that changes
Event-scale procurement is not a bigger version of the household problem. The variety constraint, which is nearly always slack for an individual, becomes the binding one, and the optimum moves accordingly.
— Scaling Household Purchasing Algorithms to Event Logistics (Journal of Applied Consumer Logistics, 2024)
Learning objectives#
By the end of this lecture you should be able to:
- apply week 10’s formulation at fifty-person scale
- explain why the variety constraint binds harder here
- say why a single-SKU-optimal inventory is the wrong answer
- treat event scale as harder, not merely larger
What carries over unchanged#
- the standard-drink conversion from week 2
- the price decomposition from weeks 1, 3, 4 and 6
- the dynamic-programming formulation from week 10
- the externality framework from week 11
The method is the method. Only which constraint binds has moved.
Variety stops being slack#
Why the constraint moves#
For one shopper:
- the best-index product, bought repeatedly, is usually an acceptable basket
- so variety is slack, and greedy often coincides with optimal
For fifty people:
- that same single-SKU inventory is trivial to construct and reliably wrong
- so variety binds, and greedy stops coinciding
What does not scale away#
Week 11’s externality does not disappear at event scale. It is multiplied by the guest count.
That multiplication is why the reading treats event purchasing as a harder problem rather than a bigger one, and why your externality statement has a specific quantity available to it that a shopper-scale statement does not.
Back to week 1#
The semester opened on what the liquid costs to produce, before any tax, margin or placement.
Twelve weeks later you are asked what a fifty-person inventory costs — and you can now decompose that number into the same four terms, defend each one, and say what the answer leaves out.
Common errors#
- presenting method for eight minutes and the answer for two
- a spreadsheet whose totals disagree with the slides, which is the most common defect and undermines everything else
- an externality statement that forbids no inventory
- an optimum that cannot physically be carried to the venue
Where this connects#
- the week 12 lab is the dry run, with the room asking the three standing questions
- the Event Portfolio is due at the close of the teaching period, not the end of this week.