When Your Recipe Becomes a BOM: Scaling Food Production

Your recipe card says 40 units. It lists flour, sugar, butter, and a note about oven temperature, and for two years it has been exactly enough. Then a regional grocery chain orders 900 units a week and asks for a lot code on every case, a best-before date printed rather than stickered, and a spec sheet listing every allergen in the facility.

That card is no longer a recipe. It's a bill of materials, and it's missing about half its lines. Nothing about your product changed. What changed is that someone downstream now needs to know which batch a specific case came from, and your system for answering that question is a whiteboard and your production lead's memory.

This is the transition most Canadian food brands hit somewhere between the commissary kitchen and their first real production run. Here's what actually has to exist before that order ships.

The Moment a Recipe Stops Being a Recipe

A recipe answers one question: how do I make this. A bill of materials answers a harder set of questions, and it's the difference between cooking and manufacturing.

What a BOM carries that a recipe doesn't

• Every component, not just food. The 500 mL container, the lid, the sleeve, the label, the case, the pallet wrap. If it ships with the product, it's a BOM line with a cost and a quantity per unit.

• Yield, stated as a number. Not "makes about 40." A stated theoretical yield you can measure actual output against.

• Lot linkage. Which incoming ingredient lots went into which finished batch, and which finished lots went into which shipment.

• Cost per unit at the case level, including packaging and labour, because that's the number a buyer's price negotiation will test.

• Allergen and changeover data, because running a sesame product before a sesame-free one has sanitation implications your recipe card never had to hold.

The practical trigger point is usually a specific request from a specific customer. A grocery buyer asks for a lot code. A distributor asks for a product spec sheet. An insurer asks how fast you can identify affected product. Each of those is a question about data, not about food, and most kitchens discover in that moment that the data exists but lives in four places: a notebook, a spreadsheet, the packaging supplier's order history, and someone's head.

Scaling doesn't break your recipe. It breaks the assumption that one person can hold the whole picture.

The version problem nobody plans for

There's a second thing a BOM does that matters more than it sounds: it has versions. When you switch to a different supplier's cocoa, adjust a fill weight, or move from a 473 mL cup to a 500 mL one, you've created a new version of the product, and every batch made after that date belongs to it. A recipe card gets edited in place, and the old numbers vanish. A versioned BOM keeps both, which is the only way to answer "what exactly was in the product we shipped in March" three months after the fact. That question arrives from customers, auditors, and occasionally lawyers, and "we changed something around then" is not an answer any of them accept.

What the Systems Conversation Looks Like at This Stage

There's a natural ladder here, and skipping rungs costs money in both directions. Spreadsheets handle a single product line and a handful of ingredient lots fine. An inventory app handles more SKUs but usually treats production as a single event rather than a process with inputs, outputs, and variance. Production and ERP platforms handle batches, BOMs, and lot genealogy properly, and they ask for structure and discipline in return.

Where you sit on that ladder should be decided by your traceability obligations and your SKU count, not by how ambitious the software sounds. Once you've written down what your customers and regulators actually require, a roundup of the best ERP for manufacturing is worth scanning to see which platforms are built for batch production versus discrete assembly, because food-specific features like recipe versioning, catch weights, and expiry-driven picking either exist or they don't.

One piece of context that should shape how you read any vendor list: Canada's food and beverage processing sector is the country's largest manufacturing industry by both GDP and employment, and roughly 91% of processors employ fewer than 100 people. The typical Canadian food manufacturer is a small business. Most software marketing is written for the other 9%, which is why capability lists tend to describe multi-plant scheduling problems you don't have while glossing over the lot-tracking basics you do.

Buy for your next two years of orders. If the tool can't be run by the people you actually employ, its feature list is irrelevant.

Traceability Is the Requirement That Sets Your Deadline

This is the part that isn't optional and isn't negotiable, so it should drive the timeline rather than follow it.

Under the Safe Food for Canadians Regulations, traceability means knowing one step back and one step forward: who you received a food from, and who you provided it to. The obligation that catches people off guard is the response window. Traceability documents must be made available to the CFIA within 24 hours of a request, in English or French, and sooner if there's believed to be a risk of injury to human health.

Twenty-four hours is a systems requirement disguised as a paperwork requirement. It means the link between incoming ingredient lots, production batches, and outbound shipments has to be recorded as it happens, at three points:

1. Receiving. Supplier, product, lot number, quantity, date. Every incoming ingredient and every packaging component.

2. Transformation. Which received lots went into which batch, with the batch's own lot code and date.

3. Shipping. Which finished lots went to which customer, in what quantity, on what date.

Miss any one of those three and the chain has a gap that no amount of weekend reconstruction closes cleanly. CFIA recall statistics have run in the range of 220 to 250 recalls per fiscal year over recent years, and allergens remain one of the most common reasons. The relevant question isn't whether you'll ever be involved. It's how long it would take you to identify exactly which cases are affected and who has them.

Your lot code is the key to all of it, so give it structure rather than a sequential number. A code that encodes production date, line or shift, and product version lets a person read useful information off a case in a warehouse without a database lookup, which matters at exactly the moment when speed counts. Keep the format consistent across every SKU, document what each segment means, and make sure the same code appears on the production record, the case label, and the shipping document. Three different numbering habits across those three places is the most common reason a traceability exercise stalls.

Run the test before anyone makes you. Pick a finished case from last month, and see how long it takes to name every ingredient lot in it and every customer who received that batch. If the answer is measured in days, you have your answer about what to fix first.

Packaging Belongs in the BOM

Packaging is the component category that gets left out of production planning most often, usually because it's ordered by a different person on a different cycle than ingredients. That works until the first time a run stops because the lids arrived and the sleeves didn't.

The mismatch is structural. Ingredients are bought in units that scale smoothly with batch size. Packaging is bought in cartons and case packs with minimum order quantities and lead times that have nothing to do with your production calendar. A 900-unit run needs 900 containers, 900 lids, and 900 labels, and if your label MOQ is 5,000 with a three-week lead time, your production schedule is now a procurement schedule.

The fix isn't complicated, it's just unglamorous: forecast packaging off the production plan rather than off the last time you ran low. If you know your run schedule eight weeks out and your BOM states quantity per unit, required packaging quantities are arithmetic. Add the supplier's lead time and MOQ to each line, set a reorder point that accounts for both, and packaging stops being the thing that decides when you can produce.

The third row causes more grief than its size suggests. When an ingredient statement changes, the old label stock doesn't disappear, and a warehouse under pressure will use what's in front of it. Versioning printed materials inside the same system that holds your BOM is what prevents an obsolete allergen statement from shipping.

Yield, Waste, and the Number You Quote

Theoretical yield is what the recipe says. Actual yield is what came off the line. The gap between them is where margin goes, and until you record both, you're quoting prices against a cost you've estimated rather than measured.

Track per batch, every batch:

• Theoretical versus actual finished units

• Ingredient quantities issued versus BOM-standard quantities

• Packaging consumed, including damaged and misprinted units

• Rework and product held or destroyed

• Labour hours attributed to the run

Two or three months of that data will tell you things a spreadsheet forecast can't. Maybe your 8% assumed waste is really 14% on one product and 3% on another. Maybe your Friday runs yield consistently lower than your Tuesday ones, which is a staffing pattern rather than a recipe problem. Maybe the SKU you priced most aggressively is the one carrying the most packaging cost per unit sold.

This is also the data a co-packer conversation depends on. If you eventually outsource production, you'll be asked for a documented BOM, a stated yield, and a spec sheet. Companies that have been tracking properly hand it over. Companies that haven't spend six weeks reverse-engineering their own product.

Four Things to Do Before You Buy Anything

1. Write the BOM out fully for your top three SKUs, packaging included, with quantity per unit and current cost per line. If you can't complete it, that's the gap, and no software fills it for you.

2. Run a mock recall on one finished lot and time it honestly. That number sets your urgency.

3. Count your SKUs and your ingredient lots per month. Under roughly 20 SKUs with simple inputs, a disciplined spreadsheet plus a real receiving log can still hold. Past that, manual tracking fails quietly rather than loudly.

4. Name the person who will own the system. Not the department. The person, with hours protected for it. This single decision predicts whether an implementation works better than the choice of platform does.

The Short Version

The shift from kitchen to production isn't really a shift in equipment. It's the point where your product data has to leave people's heads and live somewhere queryable, because customers, auditors, and regulators start asking questions that only records can answer.

Start with the BOM for your top SKUs and put packaging on it. Time a mock recall against the 24-hour window the SFCR sets. Then choose tools that fit the size of business you actually run, and give one named person the job of running them.