Planning production so you stop throwing food away

The bin at the end of the day is the most expensive thing in the kitchen, and it is the only cost nobody writes down.

That is what makes it hard to fix. It does not arrive as a bill. It leaves as a bag, quietly, at closing, and by the next morning the only trace is a vague sense that the week was busier than the takings suggest. All figures below are ratios, so substitute your own currency.

What a thrown-away item actually costs

Not its ingredients. Its full cost.

By the time something goes in the bin you have already spent the ingredients, the labour, the oven time, the share of your rent for that hour and often the packaging too, because it was boxed and labelled and ready. None of it is recoverable. Value waste at the total cost of making the thing, the same number you price from, not at the cost of the flour. If you have never assembled that total, the full cost breakdown for one order is the same arithmetic read from the other end.

Do not value it at retail price. Tempting, because it makes a dramatic number, and wrong: you never had that sale, so you never lost the margin. Overstating waste pushes you into underproducing, which has its own cost and is much harder to see.

Count the lost margin separately when there was a customer. An item destroyed after it was sold costs the full cost plus the sale. Different event, different fix.

How much to make: the rule that decides one more unit

Making one more unit is a bet. If it sells you gain the margin, price minus cost. If it does not you lose the full cost. So it is worth making when the chance it sells, times the margin, beats the chance it does not, times the cost. Rearrange that and almost everything cancels:

Which is the same as saying: at least 1 minus your gross margin.

At a 40 percent margin you need to be 60 percent confident of selling the next one. At a 70 percent margin, only 30 percent confident. That explains something which looks like recklessness from outside: a high-margin product should be made speculatively and a low-margin one should not. The cafe that always has pastries at 4pm and the bakery that sells out of sourdough by 10am can both be right.

Now the part that changes the answer most. If unsold units are not a total loss, the threshold collapses:

Take a product costing 60 to make and selling at 100. Full loss on anything unsold means you need 60 percent confidence. Add a day-old channel that recovers 50 and the threshold drops to 20 percent. Same product, same forecast, completely different production decision, purely because of what happens to leftovers.

So a discount shelf, a staff price or a second-day wholesale account is not a way of tidying up mistakes. It is a production planning tool, and setting one up will change what you make tomorrow more than any forecasting will. One condition: it must not cannibalise full-price sales. Day-olds at the end of the day are fine. Day-olds at 9am train your regulars to wait.

Before setting one up, work out what it would genuinely recover, because salvage value is not the discount price. It is the discount price minus what the salvage itself costs you. A shelf at the front counter recovers nearly all of what it sells for, because the item is already made, boxed and standing there. A route that involves unwrapping, re-cutting, re-labelling and driving somewhere recovers much less, and once the handling time is counted it can recover nothing at all. Work out the net figure, put it into the second formula above and see where your threshold lands. A route that recovers most of full price moves it a long way. A route that recovers a sliver barely moves it, and may still be worth running for other reasons, just not this one.

Cannibalisation is the failure to watch for afterwards, and it is easy to miss because total units sold will look fine while the mix rots underneath it. It shows up as a fall in full-price units, not in the total. So count full-price sales on their own for a few weeks after you start. If that count drops, the fix is almost always timing and visibility rather than price: later in the day, away from the main counter and never announced in advance.

Batching: the quantum you cannot get away from

Ovens, mixers, trays and proving space all impose a minimum. If your tray holds 24 and you need 9, the other 15 are a bet you did not choose to place, and the rule above cannot help because you were never offered the choice. Three ways out, in order of how often they work.

Shrink the quantum. Smaller trays, half batches, a second tin size. Costs a little efficiency per unit, buys back a lot of flexibility and for most small operations it is the right trade.

Move it upstream. Batch the thing that keeps, not the thing that does not. A large batch of sponge that freezes is not a bet. A large batch of finished, filled, decorated cake is.

Sell the surplus on purpose. If the quantum is genuinely fixed, plan the overage into a bundle, a market stall, a predictable wholesale drop. Surplus you planned for is stock. Surplus you did not is waste.

Make ahead or make to order: split by stage, not by product

The usual framing is wrong. "Do I make this to order" is not a question about the product, it is a question about each stage of it, because the stages have wildly different shelf lives. Sponge freezes for weeks. Buttercream keeps refrigerated for days. An assembled, crumb-coated cake keeps a couple of days. A finished decorated cake keeps until it stops looking finished, which is often shorter than its food safety window.

So: make ahead at the longest-keeping stage, and no further. Bake sponges in big batches and freeze them. Make buttercream to match your week, not your day. Assemble and decorate to order. That one change is what lets a small kitchen absorb a busy weekend without working through the night or binning Sunday's leftovers.

Do that stage by stage across the menu and you find the real bottleneck, which is rarely the one you would have named. Seldom the oven. Usually decorating, fridge space or the two hours a filling needs to set before anything else can start.

Finding yours does not need a study. A bottleneck is wherever work piles up waiting: the trays stacked beside the fridge because there is no shelf free, the row of assembled cakes waiting on one pair of hands, the two hours in which nothing can move because a filling has to set. Walk your own kitchen halfway through the busiest day of the week and look for the pile rather than for the machine. Whatever the pile is sitting in front of is the constraint, and adding capacity anywhere else buys you nothing.

Then write the stage split for your three highest-volume products, in three columns: what freezes, what keeps for days under refrigeration and what has to happen on the day. Do it honestly, testing where you are unsure rather than assuming the cautious answer. Most products end up with more in the first two columns than their maker expected, and every hour moved leftwards out of the third column is an hour bought back on the day it is scarcest.

Portioning and the yield gap

Anything sold in portions has a gap between the portions you planned and the ones you can sell. A tray planned at 20 yields 18, because of the ends, the one that broke and the two that came out uneven. That 10 percent belongs in the price of the portion, not in a vague sense that traybakes are less profitable than they look. Find it by recording both numbers every time: what went in, and what came out sellable. Record only one and you will never see the gap, which does not stay constant either, because it moves with who is cutting and how tired they are.

Two things shrink it. Cut to a marked template rather than by eye, so the same tray always yields the same count. And decide in advance what happens to offcuts, because offcuts with a destination are a product and offcuts without one are waste.

The three buckets, and why they need different fixes

Recording waste as one number is how it stays unfixed. Three kinds, almost unrelated to each other:

• Production loss. Trim, offcuts, failed bakes, the batch that split. Fixed by process and portioning discipline. Should be small and steady, and a spike means something changed in the method or the equipment.
• Unsold finished goods. Fixed by the arithmetic above and by a salvage route. A forecasting and pricing problem.
• Ingredient expiry. Fixed by purchasing, not production. A bigger sack at a better unit price is only cheaper if you use the sack. Throw away 15 percent of a bulk pack and a 10 percent bulk discount lost you money.

Log five fields, no more, or nobody will do it: date, item, quantity, reason, full cost. Keep the reason list short, because a long one means everything gets tagged "other".

Watch waste cost as a share of production cost, per product

Not waste as a percentage of revenue. That is the number that makes waste look trivial, because revenue is the biggest figure on the page. Watch waste cost as a percentage of production cost, per product, per week. Per product is what makes it actionable, because a total is always "a bit high" and never points at anything. Weekly is what makes it visible, because a month hides a bad Tuesday. Most products will sit in a stable band. The one that does not is the one to work on.

Do not go looking for a target figure to aim at. Any benchmark you are offered is measuring a different menu, a different shelf life and a different definition of what counts as waste, and chasing someone else's number will have you underproducing or ignoring a real problem depending on which way the mismatch falls. Your baseline is your own first four weeks.

So record it per product, leave it alone for a month and then read the spread rather than the level. The signal worth acting on is a product sitting well outside its own band, or a band drifting upwards week after week. Both point at something specific you can go and change. A single headline percentage, however respectable it looks, points at nothing.

What to do this week

• Write down cost divided by price for your three highest-volume products. Those numbers tell you how speculatively each one should be made.
• Ask whether you have a salvage route. If not, that is the highest-value thing on this list.
• Write the stage split for one product: what freezes, what keeps, what is made the same day.
• Count one tray in and one tray out, and see whether your yield is what you thought.

What makes waste data worth having is valuing it at full cost and keeping it per product. That is the decision that matters, and ibakepro makes it the default: a waste entry is logged against the product at its full made cost, including labour and overhead, precisely because ingredient cost understates it by the largest part.

The goal is not zero waste. Zero waste means you sold out early every day and turned people away, which costs more than the bin does. The goal is knowing what the bin costs, and choosing it.