Lulu Hypermarket

Lulu Retail Holdings

AI Operating System

DeployOne
Viewing as

Operating scope

Country head

Capability

All capabilities

Active handovers

0

Outcomes written back

0

Information density is explain.

MarginOS · screen 03 of 07

Waste and fresh yield

Waste that tracks production timing is not a forecasting problem.

3 issues · 2 waiting on you

Issues in scope

3

3 instances across 2 places

Waiting on you

2

0 already decided in this session

Three largest

AED 26.6m–AED 80.3m

Modelled annual value, three largest only

Running without asking

0

1 armed and waiting on a threshold

Decision owner

Country head

Returns to Country head, Group

Scope

Oman

27 stores · all capabilities

Listen

Fresh waste by storeForecast errorDelivery timingCentral kitchen output

Process

Waste against forecast errorDivergence means the cause is not demandKitchen timing against the demand curve

Act

Production timing proposalTo the country headStanding arm on divergence
Written backEvery outcome returns to the systems that produced the evidence.
KindOrder

Queue

3 issues · 2 waiting on you
Armed

Nothing currently notices when waste stops tracking demand

Jaber Al Ahmad
What happened
Waste is reported as a percentage. Whether it is moving with forecast error or away from it is not measured anywhere, so a cause outside the store is indistinguishable from a cause inside it.
Why it matters
Divergence is the single most useful signal in fresh, because it tells you the store is not the problem. Without it, every waste conversation starts by blaming the store, which is where most of them also end.
Already done
Armed on divergence rather than on the waste level, so it fires on cause rather than on symptom.
Worth
AED 15.4m to AED 47.5m a yearModelledDivergence separates a kitchen cause from a store cause. Value is the fresh conversations that stop starting by blaming the store.Waste causes correctly attributed
Then what
Threshold armed on divergence across every storeGroupStanding, no return date
Threshold
Waste and forecast error diverging in any store beyond tolerance
Fires
Fires with the store and the likely external cause attached, because divergence means the cause sits outside the store
GroupMarginOSDeployOnePredictDeployOneBrain
Approve

The stores affected cannot fix this themselves

Oman
What happened
A central kitchen window serves multiple stores. Four store managers receiving a waste action can do nothing about when production runs.
Why it matters
Sent to stores this produces effort without a change, and it takes attention from the actions stores can act on. The decision sits with the country because the asset does.
Already done
The proposal is prepared at country level with the store level evidence attached.
Worth
AED 6.3m to AED 18.2m a yearModelledA proposal sent to the owner of the asset rather than to stores that cannot change it. Value is the share that gets executed rather than acknowledged.Waste actions reaching the right owner
Then what
Production proposal with timing evidenceCountry headReports back 30 Sept
If approved
The production proposal goes to the Oman country head with the timing evidence
If declined
It goes to store managers who cannot change a central kitchen window
Country headMarginOSDeployOnePredictDeployOneBrain
Choose

Waste at 5 Oman stores follows the kitchen production window, not forecast error

Oman
What happened
At these stores waste rises and falls with when the central kitchen produces, while forecast error stays flat. Everywhere else in the country the two move together, which is what a demand problem looks like.
Why it matters
Fresh waste is treated as a forecasting failure almost everywhere, and forecasting teams have been working on it. At these stores forecasting is not the problem and no amount of forecast improvement will move the number.
Already done
The divergence is isolated and three fixes are costed, each landing on a different owner.
Worth
AED 4.9m to AED 14.6m a yearModelledWaste tracking production timing rather than forecast error. Value is the share of fresh waste attributable to timing, times what a window change recovers.Waste against forecast error divergence
Then what
Chosen fix costed and scoped to the kitchenCountry headReports back 30 Sept
Country headMarginOSDeployOnePredictDeployOneBrain