Why a roastery can't roast ahead to get the week calm

The day the week is built backwards from

Start with the one thing on the week that cannot move: a café expects its beans on a set day. Not "sometime this week." A day. The bar has service to run and a hopper it is nearly through, and whoever runs it planned around you.

So the week gets built backwards from that delivery. If the beans land Thursday, the batch has to come off the roaster before then, with time to rest. Which means the order had to be confirmed earlier still, and the green for it pulled before that. Roasters describe the rhythm plainly: customer commits Monday, roasted Wednesday, delivered Thursday. One fixed point, and three earlier steps hanging off it.

That is the wholesale clock. There is a second one, and it does not keep the same time.

Two clocks, one roaster

The café order is a commitment to another business with a deadline attached. Miss it and a bar opens without the coffee it sells. That side is firm, or it needs to be.

The shelf is different. Retail bags on the counter, the orders that come in through the website: those are a guess at demand you make in advance and find out about later. Nobody committed. You are reading the last few weeks and betting on the next one.

Both of those draw on the same two things: hours on the one roaster, and the same sacks of green behind it. A morning spent roasting to hit a café's Thursday is a morning not spent building the counter's stock, and the other way round. The week is those two demands negotiating over one machine.

Why you cannot roast Monday and coast

The obvious fix is to roast everything early and spend the rest of the week calm. Coffee will not let you.

Here is how roasters talk about it, so correct me if the trade says otherwise: roasted coffee stays good for a matter of weeks, and it starts flattening well before that. A café tastes the difference. Roast the whole week's worth on Monday and by the time the last of it reaches a bar it is past its best, and the buyer notices in the cup even when they cannot name why. Roasting ahead does not remove the planning problem. It swaps it for a staleness problem, and the staleness problem is the one your customer meets.

So the calm is not available. The freshness window is the reason the week has to be shaped the way it is, and it is the reason you cannot plan your way out of the timing by working a long Monday.

Roast-to-order or roast-to-stock, decided every week

This is where the two clocks turn into an actual decision, not a philosophy. Every week you are splitting the roaster's hours between coffee that has a name on it and coffee that does not.

Roast-to-order is the café work: a firm order, roasted close to delivery, out the door fresh. Roast-to-stock is the speculative bagging for the counter and the web, sized to a few days of expected sales because beyond that it drifts out of the freshness window. Roasters I have read put that number at something like five to seven days of sales per batch, for exactly that reason.

The decision is which accounts sit on which side, and how much you dare bag ahead. Get the stock guess high and it sits, flattens, and you either sell it cheap or eat it. Get it low and the counter is bare on a good Saturday. Neither miss is free.

Where the wholesale number even comes from

Ask a roaster where this week's café figure comes from and the honest answer is often: the standing orders, plus whatever changed since. A regular said hold the dark this week. Someone added a second bag. One went quiet. The "standing" order is a starting point you then edit from memory.

I know that editing-from-memory feeling from the other end of it. I grew microgreens and sold them to trade, chefs and restaurants, somewhere between five and ten accounts, and I planned production from scattered orders plus memory. The standing orders rarely stood. I got it wrong in both directions: waste some weeks, shortfall others. A highly perishable thing on a narrow clock, promised to kitchens, planned off half-remembered messages. That is the same shape as a roaster's week, even though I have never roasted a bean.

The part worth fixing is the wholesale side, because it is the part that is actually knowable. A late change or an extra café order reshuffles the roast schedule and the green you pull, and if the only place that order lives is a WhatsApp thread and your head, the reshuffle happens twice: once when it comes in, once when you forget it did. Getting the café commitments off memory and onto one settled list is not about forecasting anything. It is about the roast plan being built from what was actually agreed, which is roughly what a production plan you will keep up to date is for. That much is a solved problem, and it is where something like makeweek earns its place: the wholesale orders in one list instead of six threads. It does not size your batches or read the counter for you. It gets the agreed part out of memory.

The green sitting behind all of it

Underneath the roasting there is a slower clock that changes where the guessing lives. Green coffee keeps, in the trade's telling, somewhere between six months and a year, against the few weeks you get once it is roasted.

That gap matters. It means the real bet is not what you roast on Wednesday, it is what you committed to buy months ago. You cannot un-roast a stale bag, but you are not deciding fresh-versus-stale at the roaster. You decided the volume upstream, when you bought the green. The weekly roast is just spending down a commitment you already made. So the guessing does not disappear, it moves to the sack, where at least it keeps for months while you work out what you were right and wrong about.

Is this your Tuesday?

The point of walking the week like this is to find out whether a roaster reads it and nods, or reads it and says no, you have the order of things wrong.

So, two questions, and they are genuine ones. Where does your roast-to-order line sit right now: which accounts get roasted close to delivery, and how much do you bag on spec for the counter and hope moves? And when a café changes its order on a Tuesday, where does that change live between then and the roast, so it does not get lost?

If you supply a handful of cafés, you may be holding all of this in your head and mostly getting away with it. Add a few more accounts and the standing orders that never stand start colliding, and the roaster's hours stop stretching. That is the point where the café side wants to be a list you can trust rather than a memory you keep re-checking. Tell me where I have the roastery week wrong. It is your trade, not mine, and the probe only works if you correct it.