5am, the ovens are already running, and the morning’s production count is still mostly a guess — a bit more than last Tuesday because it’s supposed to rain later, a bit less than last Friday because the wholesale order shrank for no clear reason. By 10am, the croissant case is either half-full of things that won’t sell by close, or empty with three customers who just walked out. Both outcomes cost money. Neither one is a staffing problem.
Why this guess is harder than it looks
Most baked goods have a 24-to-48-hour sell window — bake too much and the surplus is trash by tomorrow morning, bake too little and you’re turning away paying customers before lunch. Unlike a restaurant holding shelf-stable inventory, a bakery re-runs this exact bet every single day, and the two failure modes pull in opposite directions: the fix for stockouts is baking more, which is the direct cause of the waste you’re trying to avoid. Without real numbers behind the decision, a kitchen just oscillates between the two, guessing a little differently each morning and getting burned by whichever guess is wrong that day.
What actually needs to happen
Replace the guess with what actually sold
Real demand data — what actually sold, by item, by day of week, by whether there was a wholesale order that morning — replaces the gut-feel adjustment with a number a baker can actually plan against. Asky’s stock tracking shows what’s moving before the case is empty or the tray is cold, so tomorrow’s production count is based on this month’s Tuesdays, not a feeling about today’s.
Stop re-guessing the orders that never change
Some of that daily count isn’t a guess at all — it’s a standing wholesale account that orders “48 croissants every Tuesday and Friday” without fail. Re-typing that same ticket every week is where a second kind of error creeps in: a repeat order that reorders itself removes the guesswork entirely for the volume that was never actually in question, and frees up the morning to focus on the part of production that genuinely does swing week to week.
Forecast the spikes months out, not days out
Thanksgiving pies, Christmas orders, Valentine’s Day, Mother’s Day — each one is a single massive day that has to be planned months, not days, ahead. Getting that forecast wrong either turns away holiday orders at the worst possible moment or leaves a walk-in cooler full of product nobody’s coming back for. The same demand data that tunes a normal Tuesday is what makes a realistic call on how much bigger this Thanksgiving needs to be than last year’s.
Make the plan simple enough for whoever’s actually on the overnight shift
The industry is projecting 53,500 unfilled US baking jobs by 2030, and overnight and early-morning shifts are the first thing new hires balk at. A production plan that only works if one specific veteran baker is on shift isn’t really a plan — it’s a single point of failure. Real demand numbers, laid out clearly instead of buried in one person’s head, mean whoever’s actually on the overnight shift tonight can follow the count without having to guess at what “feels right,” the same way the owner would.
Make the waste itself visible, not just the sales
Waste is easy to lose track of when it happens quietly in a bin at closing instead of on a screen anyone actually reviews. Folding what didn’t sell into the same end-of-day reporting as what did — instead of a mental note that’s forgotten by the next morning’s rush — means a manager can see whether last week’s adjustment actually worked, item by item, instead of guessing at whether things “feel” a little better.
Closing
The goal was never zero waste — a bakery that never throws anything away is a bakery that’s turned away customers all morning to get there. The goal is closing the gap between what the ovens made and what the day actually needed, a little more each week. Every tray that goes in the trash and every customer who walks out empty-handed is the same mistake, made in opposite directions.