How Contract Caterers Should Present Forecast Error, Food Waste and
How Contract Caterers Should Present Forecast Error, Food Waste and Satisfaction Hypotheses in a Monthly Operating Review
A monthly operating review is where a contract catering account either earns or loses credibility. The numbers — covers, waste weight, complaint counts, satisfaction scores, menu performance — rarely speak for themselves. What clients such as facilities managers and procurement leads actually need is a reasoned chain that connects them: a forecast error led to a production decision, which led to surplus or shortage, which showed up in waste and complaints, which moved satisfaction, which
Why Hypotheses Instead of Conclusions
The temptation in a monthly review is to narrate: waste went up, satisfaction went down, so the menu was changed. That narrative may be true, but it does not survive scrutiny. A client will reasonably ask whether the waste increase was forecast-driven or recipe-driven, whether the satisfaction dip was caused by availability or by something else entirely, and whether the menu change will actually fix anything.
Framing the relationships as hypotheses changes the tone of the meeting. A hypothesis is stated in advance, checked against evidence, and either supported, weakened or revised. It gives both sides a shared method:
The caterer can show analytical discipline rather than defensiveness.
The client can challenge the reasoning, not just the outcome.
Actions become experiments with a defined next-month test, not vague commitments.
The core logic worth presenting on one slide is a chain:
Forecast → production decision → availability or surplus → diner experience → menu or process intervention → next-month result.
Everything in the review should map back to a link in that chain.
The Five Variables and How They Connect
Demand-Forecast Error
Forecast error should be reported at two levels, and the distinction matters more than most clients realise.
Total-demand error: did the kitchen produce roughly the right number of meals overall?
Item-level or menu-mix error: was production distributed across dishes the way demand actually arrived?
A kitchen can hit total covers almost exactly while over-producing one dish by 25% and selling out of another. Aggregate accuracy can coexist with item-level failure, and item-level failure is what generates waste and shortages at the same time.
Present forecast error by service period and by menu item where possible. A single site-wide percentage hides the patterns that drive decisions.
Food Waste
Waste is the physical evidence of forecast and production decisions. Break it down rather than reporting one number:
Preparation waste (trim, spoilage)
Overproduction and unsold food
Plate waste returned from customers
Segment by day of week and service period. Waste that climbs every Friday points at a production plan that ignores actual attendance. Waste concentrated in one dish points at demand, portioning or recipe issues — which is a different hypothesis with a different fix.
Packaging belongs in this section as a measurable factor, not a footnote. Portion consistency depends on standardised container sizes: when kitchens portion into calibrated bowls and boxes — a 750ml kraft bowl versus a 1000ml one — serving sizes stop drifting, and portion-driven waste becomes easier to attribute and control. The container is also where plate waste becomes visible: a clear record of what comes back in a standard-format container is easier to quantify than scraps on mixed tableware.
Complaints
Complaints are not a single category. Separate them by type, because each type points at a different link in the chain:
Availability complaints ("the dish I wanted had sold out") → under-forecasting or mix error.
Quality complaints (temperature, freshness, texture) → holding times, equipment or production scheduling.
Experience complaints (queues, spills, packaging failures) → service design and materials.
The third group is worth isolating because it is often the cheapest to fix. A leak or spill incident traced to an unsuitable container is a packaging-selection issue, not a kitchen-performance issue. Recording the container used for each spill or leak incident lets the review show whether incidents track with one product format — and whether switching to a fitted-lid format, such as lidded rectangular containers for saucier dishes, removes the problem. That is exactly the kind of low-cost intervention a hypothesis framework is built to surface.
Satisfaction
Satisfaction scores should be cut against operational conditions, not just averaged. The most revealing comparison in many reviews is:
Satisfaction on days when popular dishes were available all service vs. days with stockouts.
Satisfaction before and after a menu change.
Satisfaction by shift or service period.
If satisfaction on stockout days runs visibly below other days, that is an association worth testing — it does not prove causation, but it justifies an availability intervention and a defined next-month check.
Menu Changes
Every menu change should be logged as an intervention with a hypothesis behind it: what problem it addressed, what evidence prompted it, and what result would count as success. A change that reduces waste by a visible margin while complaints stay flat and satisfaction holds is a candidate for retention and confirmation over two or three cycles. A change that trades one problem for another should be revised, not defended.
A Workable Review Format: Hypothesis, Evidence, Decision, Test
One table, applied consistently month after month, does more for review quality than any dashboard redesign. Columns:
Column
Content
ID
A stable reference (H1, H2…) carried across months
Hypothesis
One sentence, falsifiable: "Over-forecasting on stable dishes drives production waste"
Evidence this month
The actual numbers: forecast variance, waste weight, complaint counts, satisfaction split
The measurable condition that will confirm or refute the action
Three rules keep the table honest:
No hypothesis without a next-month test. An action with no defined check is an opinion.
One variable per hypothesis where possible. "Waste rose because of the menu change and the weather" cannot be tested.
Report the refutations. A review that only shows winning hypotheses stops being trusted by month three.
Worked Example for a Single Account
A composite example of how one month might read:
H1 — Over-forecasting drives waste. Forecast ran 12% above actual covers; production waste rose 9%. Interpretation: consistent. Action: tighten the forecast on the three stable, high-volume dishes. Test: waste falls without availability complaints rising.
H2 — Mix error drives simultaneous waste and stockouts. Total covers accurate within 2%, but one hot dish held 20% surplus while a second sold out by 12:40. Interpretation: item-level error, not total-demand error. Action: reallocate production between the two dishes. Test: item-level forecast error narrows.
H3 — Availability affects satisfaction. Satisfaction averaged 3.8/5 on stockout days versus 4.4/5 otherwise. Interpretation: association worth testing. Action: protect production of the top-selling dish through peak. Test: the gap narrows next month.
H4 — Spill complaints track with one container format. Four of five spill complaints involved the same loose-lid container used for a saucy dish. Interpretation: packaging-selection issue. Action: move that dish to a fitted-lid container and recheck. Test: spill complaints fall to zero over the next two cycles.
H5 — Last month's menu change worked. Waste down 15%, complaints flat, satisfaction up 0.1. Interpretation: appears successful. Action: retain. Test: confirm over two more cycles.
Ten minutes of discussion around a table like this answers the questions a client actually brings: what happened, why, what was done, and how we will know it worked.
What Belongs in the Pack Beyond the Table
The hypothesis table is the spine; the review pack around it should stay lean:
Trend views for forecast variance, waste by stream, complaints by type, satisfaction — three to six months, not twelve lines of daily detail.
Cost-per-meal context, with packaging as a visible line item. Standardised containers and closures are procured, not free, and treating them as an accounted line lets the review show trade-offs honestly — a slightly higher-specification lidded container that eliminates spill complaints and remakes is usually the cheaper option once rework and goodwill are counted.
Waste-separation support: if the client has diversion targets, show how waste is separated at the service line, and where container choice (single-material formats over mixed-material ones where the menu allows) supports clean segregation without the caterer making environmental claims the data cannot yet back.
Compliance confirmations — food safety and allergen records — as standing items, unchanged unless something moved.
Common Presentation Mistakes
Correlation narrated as causation. Satisfaction fell the same month a menu changed; that is association until a mechanism and a control comparison are shown.
Hiding the misses. Refuted hypotheses are the most valuable rows in the table; they show the method working.
Only aggregate numbers. Total covers accurate, item mix wrong — the aggregate view conceals the exact problem the client cares about.
Actions with no tests. "We will monitor" is not a next-month test.
Treating packaging as invisible. Container choice, portion calibration and closure fit are operational variables with measurable consequences; excluding them leaves unexplained residual variance in the spill, waste and satisfaction data.
Getting Ready for Your Next Review
A monthly operating review built on explicit hypotheses does two things at once: it explains last month, and it makes next month predictable enough to manage. Caterers who run reviews this way tend to find that the small operational levers — item-level forecasting, portion calibration, container and closure selection — move the numbers faster than sweeping menu overhauls.
If you are reviewing how your catering operation handles portioning, takeaway formats or waste separation, the product specifications are worth a look. TakeawayPack manufactures and supplies kraft paper bowls and lunch boxes, rectangular and single-compartment food containers, clear PET/PP cups and matching lids, paper bags and cup sleeves, with size specifications listed per product — useful when you want portion sizes and closure fit defined by specification rather than habit. See the full range and dimensions at takeawaypack.com.
