Forecasting for Cyclic Menus is an AI-powered demand forecasting approach that enables predicting sales by dish, day, and location, adjusting production and purchasing to reduce waste and stockouts in HORECA chains. School canteens, hospitals, elderly care homes, corporate catering, all-inclusive hotels. They all share an operational trait that sets them apart from a la carte restaurants: their menus repeat in two, three or four-week cycles. This repetition should simplify purchasing and production planning. In practice, however, it creates a forecasting problem that most systems ignore entirely.
What is a cyclic menu and who uses it
A cyclic menu is a meal plan where a fixed set of dishes rotates periodically. Instead of a fixed a la carte menu chosen freely by diners, the operator defines what is served each day of the cycle. Monday of week 1 is macaroni with tomato sauce; Monday of week 2, chicken with rice; Monday of week 3, lentil stew. When the cycle completes, the rotation restarts.
This model is the norm in food service: school canteens, public and private hospitals, elderly care homes, corporate canteens and industrial catering services. It is also used by all-inclusive hotels, where the dinner buffet rotates thematically each week.
The scale is considerable. A mid-sized food service operator manages between 15 and 60 service points, each serving between 100 and 800 diners daily. Cycles are designed centrally, but execution is local. And that is where the challenge begins.
The problem of forecasting demand in cycles
In an a la carte restaurant, demand forecasting follows relatively stable temporal patterns: more covers on Friday than Tuesday, peaks on holidays, dips in January. A time series model can capture these patterns with enough historical data. But in a cyclic menu, the day's demand depends on the menu for that day, not just what day of the week it is.
If Monday of week 1 serves pasta, demand for flour, crushed tomato and cheese spikes. If Monday of week 2 offers rice, demand shifts to rice, eggs and plantain. A forecasting model that only looks at the calendar without understanding the cycle structure will produce systematically incorrect forecasts.
Variables that complicate the equation
The cyclic menu structure is only the first layer of complexity. In real food service operations, additional factors turn forecasting into a first-order challenge.
Dietary requirements and alternative menus
Each center serves multiple menu versions: the standard menu, gluten-free, lactose-free, vegetarian, halal. Each variant has different ingredients and different diner proportions. In a school with 400 students, there may be 30 special menus across 5 categories. Forecasting total daily demand requires summing all variants, not just the main dish.
Attendance variability
In food service, attendance is not constant. Fewer diners come on Fridays than Mondays. The weeks before holidays see 15-20% drops. In hospitals, bed occupancy fluctuates. In corporate canteens, long weekends and remote work days significantly reduce demand. A forecasting system that does not model absenteeism will produce recurring surpluses.
Portions by population
Serving sizes vary by the population served. A portion of lentils in a primary school weighs 250 grams; in a corporate canteen, 350 grams; in an elderly care home, 300 grams but with modified texture. The same cycle dish generates very different ingredient demands depending on the center.
Cycle changes
Every time the cyclic menu is renewed, typically when the season changes, the forecast loses its direct historical basis. New dishes have no prior data. If a beef stew with vegetables replaces a chickpea dish, the system needs to estimate demand without specific historical data for that dish.
Why traditional approaches fail
Most food service operators solve forecasting in two ways, and both generate serious problems.
The first is ignoring the cycle completely. An average daily consumption per ingredient is calculated based on previous weeks, without distinguishing which menu corresponds to each day. The result is a forecast that underestimates demand on popular dish days and overestimates it on less popular dish days.
The second is manual adjustment. The head chef or purchasing manager reviews the cycle each week and manually corrects quantities based on experience. This approach works reasonably well when executed by an experienced professional who knows each center. But it does not scale.
How Controliza forecasts demand in cyclic menus
Controliza's Forecast module incorporates cycle structure as a central variable in the forecasting model. This is not a superficial adaptation: the forecasting engine identifies each day of the cycle as an entity with its own demand profile.
Complete cycle modeling
Controliza imports the cyclic menu definition with all its dishes, variants and serving sizes. Each combination of day-cycle week-center becomes an independent forecast point.
Attendance forecasting by center and day
The model incorporates absenteeism patterns by center, day of the week, proximity to holidays and seasonality. If Fridays at the Valencia center show 12% fewer diners than Mondays, the forecast adjusts quantities automatically.
Adaptation to cycle changes
When a new cyclic menu is introduced at the start of a season, Controliza uses data from analogous dishes to generate an initial estimate. As real data from the new cycle accumulates, the model recalibrates automatically.
Integration with purchasing: the cycle as an advantage
This is where the cyclic menu goes from being a problem to a strategic advantage. If the system knows what will be served over the next four weeks at each center, purchasing planning can be anticipated with precision impossible in a la carte restaurants.
Controliza cross-references demand forecasts by dish and center with each recipe's recipe costings to calculate the exact need for each ingredient, day by day, center by center.
Why traditional planning tools fail when the menu changes but the operation does not
Most food service operators still plan cyclic menus with spreadsheets, ERP averages or static rules based on last week’s consumption. The logic seems reasonable: if the menu repeats, demand should repeat too. But repetition at menu level does not mean repetition at operational level. Attendance changes with school calendars, hospital census, hotel occupancy, local events, weather shifts and even channel mix between dine-in, takeaway and room service. The result is a planning model that looks orderly on paper but breaks in execution. You end up with excess mise en place for low-demand dishes, emergency purchases for underestimated ingredients and recurring gaps between recipe costing assumptions and real food cost.
This is where Forecasting becomes operational intelligence rather than a reporting exercise. Controliza’s AI does not just detect that a dish belongs to week 2 of a cycle. It learns how each dish behaves by day, location and service context, then adjusts the forecast using external drivers such as holidays, weather, events and occupancy. It also handles the complexity that usually destroys forecast quality: menu substitutions, atypical consumption spikes, changing diner counts and the interaction between standard and special diets. Instead of producing one generic estimate, the system generates a granular forecast that can be translated directly into production, purchasing and prep decisions.
The practical impact is immediate across the kitchen and supply chain. Teams can define more accurate defrosting plans, prepare the right quantities for each service and automate orders based on expected demand rather than intuition. Purchasing improves because delivery notes and supplier receipts can be matched against what should actually have been consumed during that point in the cycle. Traceability also becomes more useful: not just a compliance record, but a way to understand which ingredients moved, where deviations appeared and how those deviations affected waste. For operators managing recurring menus, this closes the loop between forecast, execution and control.
When forecasting reaches dish level, cyclic menus stop being a blind spot and become a source of predictability. That translates into lower waste, fewer stockouts and tighter production discipline. Controliza typically helps reduce waste by 20–30%, cut stockouts by up to 40% and keep production deviation versus demand below 10%. For HORECA chains working with cyclic menus, that improvement is not only financial. It means more stable service, better food cost control and less daily firefighting for kitchen, procurement and operations teams.
Measurable impact
Food service and catering operators implementing Controliza with cyclic menu modeling achieve tangible results in the first months of operation:
Data measured in active Controliza clients.
