40 Chicken Dum Biryanis
The four-week same-Saturday average is 40.25 portions. A kitchen may approve 42 after adding a small service buffer and rounding to its batch size.
Menu Demand and Preparation
Forecast portions at the level the kitchen can act on.
Openfluid estimates item demand by outlet, then separates the statistical forecast from the preparation quantity approved by the operator.
Item and outlet grainBuffers shown separatelyBatch and holding rules supported
Menu Demand
Kitchen logic
The forecast and the prep plan are deliberately different numbers
When these quantities are merged, teams cannot tell whether an error came from the forecast or from the operating policy.
ForecastExpected portions
+What the demand model estimates for the service period.
Operating bufferExplicit protection
→A chosen allowance for service risk, not hidden model inflation.
Preparation planKitchen-ready quantity
Rounded for batches, shelf life, holding limits and existing prepared stock.
Saturday example
One item can require three different outlet plans
The values are derived from the synthetic item-demand series in the prototype workbook and rounded for readability.
38 portions
The demand level is slightly lower at 37.5, while expected order count is higher. The mix of other menu items matters before changing production.
37 portions
The same item has a 37.25 baseline. Reusing the Jubilee Hills plan here would create avoidable excess.
30 / 27 / 25 Chicken 65
The outlet spread is different again, showing why a chain-wide percentage uplift is a weak preparation method.
Operating constraints
Preparation rules belong in the product, not in a manager’s memory
These are business rules around the forecast. They should remain visible and editable rather than being buried inside a model.
Batch size
Round to feasible production batches rather than showing a mathematically precise but unusable quantity.
Holding time
Limit early preparation for items whose quality or safety deteriorates quickly.
Existing prep
Subtract usable carried stock or pre-production already completed.
Substitution
Account for menu items that compete or substitute when one becomes unavailable.
Daypart
Split lunch and dinner when one daily total would arrive too late for the kitchen.
Override reason
Record why a manager changed the recommendation so the outcome can be reviewed later.
Decision quality
A useful screen shows the consequence of being wrong
Optimising only forecast accuracy can still produce a poor kitchen plan if the cost of under- and over-preparation is asymmetric.
Under-preparation
Stockouts and lost availability
Track when an item became unavailable, the time of the stockout and the demand that could not be served.
↔
Over-preparation
Leftovers, markdown or waste
Track prepared quantity, sold quantity, usable carry-forward and actual waste. The right buffer balances both outcomes.
FAQ
Menu demand questions.
Does POS history show preparation quantity?
Usually no. POS gives sold quantity. Preparation logs, kitchen production records or a declared baseline are needed to evaluate over-preparation.
Can the model forecast newly launched items?
A new item has limited history. Use category, recipe, price and similar-item signals, then widen uncertainty and review frequently.
Should managers be allowed to override?
Yes. The override, reason and outcome should be retained. That creates evidence about local information the model did not have.
Translate the approved menu plan into ingredients.
Recipe quantities, yield and current stock turn portions into a purchasing and transfer view.