Bakery inventory: tracking ingredient stock without counting flour every night

A shop counts what it sells. It bought 40 units, it sold 31, it holds 9, and if the shelf says 9 the books are right.

A bakery cannot do that, because nothing leaves in the shape it arrived. Flour leaves as cake. Butter leaves as buttercream that left as a wedding order three days later. Your stock does not go out of the door, it gets transformed on the way, and every transformation is a place where the count can drift.

Why bakery inventory is a different job

Three differences do all the damage.

You sell transformations, not units. To know what a Saturday consumed you have to read back through the recipes: 14 orders, each with a bill of materials, each ingredient in a different unit. Nobody does that arithmetic by hand at 6pm on a Saturday, so it does not get done, and the count goes stale at exactly the speed you are busy.

Your fastest-moving items are the cheapest. Flour, sugar and butter move constantly and cost little per kilogram, while the most expensive thing in the store room is often touched twice a month. Ranking your attention by how often you use a thing gets it backwards.

A shortage stops production, it does not just lose a sale. A shop that runs out of a line sells something else that day. A bakery that runs out of a gelling agent at 7am has a decorated cake booked for 2pm and no way to make it. The cost of the stockout is the order, not the ingredient.

The two-unit trap

This is the one that produces numbers so wrong they look like a software fault.

You buy flour in 5 kg (11 lb) sacks. You use it in grams. Those are two different units for the same item, and every stock system, spreadsheet or notebook has to reconcile them somewhere. If that reconciliation is missing, the factor applied is 1, and the arithmetic goes off by a thousand in one direction or the other.

Take a sack at 12.50 for 5 kg. Cost per gram is 0.0025. A recipe using 450 g of flour should cost 1.13 in flour.

• Stock held in kilograms, recipe written in grams, no conversion: the batch reads as consuming 450 kg. Your stock figure goes deeply negative after one bake and the flour cost in that recipe prints as 1,125.
• The same mistake the other way, stock in grams and recipe in kilograms: the batch consumes 0.45 g. Stock never moves, costs read as nothing and you find out at the stocktake.

The first is obvious within a day, because the number is absurd. The second is the dangerous one, because a cost that is slightly too low and a stock figure that barely moves both look plausible for months.

Perpetual or periodic, and which one you need

Periodic means you count on a schedule and the figure between counts is unknown. Perpetual means every movement adjusts a running figure and the count only confirms it.

Perpetual is better and it is not free: it only holds if every movement gets recorded, including the ones nobody thinks of as movements, like the 200 g taken for a test bake.

A rough decision rule. If you make fewer than about 15 orders a week from a pantry of under 30 lines, and you are the only person who touches the store room, periodic is enough: count weekly, buy from what you find. Once a second person is drawing ingredients, or once you are quoting work that depends on a specialty item you cannot get in 48 hours, periodic stops working. The gap between "I counted on Sunday" and "somebody used it on Tuesday" is where the failed Saturday comes from.

Most small bakeries end up running both: perpetual on the 20 or so lines that matter, periodic on everything else.

Par levels: the number that decides when to buy

A par level is the point at which you reorder. Set it too low and you run out during the lead time. Set it too high and your money sits on a shelf getting stale.

Use your longest realistic lead time, not the usual one: the usual one is not what hurts you. The buffer covers demand you did not forecast, and 50 percent of lead-time demand is a reasonable start.

Worked example, plain flour:

• Weekly usage: 25 kg (55 lb), read off four weeks of actual purchases.
• Longest lead time: 5 days, which is 0.71 of a week.
• Lead-time demand: 0.71 x 25 = 17.8 kg (39 lb).
• Buffer at 50 percent: 8.9 kg (20 lb).
• Par level: 26.7 kg, rounded up to 30 kg (66 lb), because flour comes in 5 kg sacks and a par level you cannot buy against is a par level you will ignore.

Then adjust by hand in two directions. Up for anything with a single supplier and no substitute, because too much icing colour is a small cost and none of it is a cancelled order. Down for anything perishable, because expiry is the form of overstock that becomes a total loss.

The stocktake that actually happens

A full count of 80 pantry lines takes two hours, gets scheduled monthly, happens twice and then stops. Design for the count you will still be doing in six months.

That last point is the whole reason to count. A corrected figure tells you what you hold today. The size and direction of the correction tells you where your process is leaking.

Where shrinkage hides

Four places, in roughly the order they cost money.

Overweighing by hand. Scooping by eye instead of weighing runs about 5 percent over. On a 450 g weight that is 22 g a batch. Forty batches a week is 900 g (about 2 lb) a week, some 47 kg (103 lb) a year, and it never appears anywhere except as a stocktake variance you blame on the count.

Spillage and breakage. Visible immediately, almost never written down. A dropped 2 kg (4.4 lb) tub that goes in the bin unrecorded is a figure your system will keep insisting you hold.

Untracked draws. Test bakes, staff samples, the handful of chocolate buttons. Individually trivial, and they are the reason a perpetual count drifts in one direction only.

Stale and expired stock. The 15 percent of a bulk pack you threw away wiped out the 10 percent bulk discount that made you buy the big pack.

Shrinkage is not a category of theft. It is the sum of everything that happened without being recorded, and the fix is a faster way to record, not a stricter rule about it.

What to do this week

• Write the stock unit next to every pantry line, and where it differs from the purchase unit, write both and the factor between them.
• Rank your lines by annual spend, mark the top 20 percent and set a par level for those lines only.
• Count them once, and record the variance rather than just the new number.

ibakepro keeps a running figure per item, deducting stock as orders move through production and flagging anything below the minimum you set, with alerts you switch on yourself. The details are in pantry inventory and stock tracking and how do I get an alert before I run out of an ingredient. The count reflects what is recorded, so it still needs a person in the store room every so often to tell it what is actually there.