Process Manufacturing: A Complete Guide to Batch and Formula-Based Production

What is process manufacturing?

Process manufacturing produces output by transforming ingredients or materials, mixing, reacting, blending, cooking, or fermenting them according to a formula, rather than assembling discrete components into an individually countable product. Water, sweetener, flavoring, and active ingredients become 1,000 liters of beverage. The next batch, run against the exact same formula, may not come out to exactly 1,000 liters.

Discrete manufacturing makes the contrast easy to see. A frame, wheels, and brakes become one bicycle, built from the same parts each time. Discrete manufacturing isn't immune to scrap, defects, rework, or yield loss, a welded frame can crack, a batch of parts can fail inspection, and material usage variance shows up there too. The difference is one of emphasis: in assembly-oriented manufacturing, product structure is often expressed through a relatively fixed component-per-unit relationship, one bicycle needs one frame. In process manufacturing, yield is often a more central part of the production model itself, because the relationship between material input and acceptable finished output can vary from batch to batch as a matter of course, not only when something goes wrong.

Hybrid process and discrete manufacturing

Many manufacturers aren't purely one or the other. A beverage company blends 10,000 liters in a single process step, then switches to a discrete packaging line to fill, label, and case individual bottles from that same batch. A nutraceutical company blends a bulk powder, then packages it into 30-serving tubs, 60-serving tubs, and single-serve packets. The process step and the discrete step describe different stages of one operation, not two separate kinds of company, and treating that combination as normal, rather than an edge case, matters for how a production system needs to be structured.

Batch vs. continuous process manufacturing

Process manufacturing itself splits into two patterns. Batch manufacturing runs in defined units of work: a specific batch size, a specific set of inputs, a start and an end, with the equipment cleaned or reset between runs. Continuous manufacturing runs material through the process in an unbroken or extended flow, refining crude oil or producing paper are typical examples, without discrete start-and-stop batches marking each unit of output.

Both are process manufacturing in the sense that matters most: material is transformed according to a formula rather than assembled from fixed components. Continuous manufacturing raises a different operational question, how a specific time window gets isolated for costing and traceability when there's no natural batch boundary to anchor it to. This guide focuses on the batch- and formula-oriented side, common across food, beverage, nutraceutical, chemical, cosmetic, and cannabis production.

How a production batch works

Take a formula for 1,000 kg of nutritional powder. It's the example this guide carries through to the end.

This defines the expected inputs for a 1,000 kg target. It doesn't guarantee the run produces exactly 1,000 kg of finished powder; the rest of this guide is largely about how that gap gets measured, planned around, and accounted for.

Formula vs. recipe vs. BOM

A formula, a recipe, and a bill of materials all answer the same underlying question, what inputs does this output require, but they're built around different assumptions, covered in full in BOM vs. formula vs. recipe. A traditional BOM, covered in full in the BOM guide, centers on a parent item and a fixed quantity of each component per unit of that parent. A formula centers on ingredients and proportions scaled to a target batch quantity, with expected yield built in from the start. Many manufacturing systems use BOM terminology to represent formulas regardless, so the conceptual distinction is what's worth holding onto, not which label a given system happens to use.

Scaling a formula

Demand for the nutritional powder comes in at 2,500 kg, not 1,000. The formula scales proportionally: divide the target by the base formula size to get a multiplier, 2,500 ÷ 1,000 = 2.5×, then apply it to every ingredient.

Proportional scaling is the foundation, but real formulas layer exceptions on top of it: an ingredient with a fixed dose regardless of batch size, a process step with a practical maximum scale, or a minimum quantity for a minor ingredient to be measurable at all. Those are exceptions to proportional scaling, not a replacement for it.

What a batch record actually is

A batch record carries a batch size, a batch number, a production date, the formula version used, the inputs actually consumed, and the output actually produced. Together those answer the two questions that matter most later, when something needs investigating: what was this made from, and what came out of it. For the 2,500 kg run, that record is what will later show the vitamin premix came in below standard potency and had to be adjusted, and what the batch actually yielded once the run finished.

Why process manufacturing gets complicated

Everything in this chapter is a variation on the same theme: the formula says what should happen, and reality doesn't fully cooperate. Five specific ways it doesn't:

Yield: theoretical vs. actual output

Three figures matter, and conflating them is a common source of bad planning. Theoretical output is what the formula's inputs would produce with zero loss. Expected output is what a run is planned to produce, given a historical or assumed yield rate. Actual output is what a specific run produced.

The 2,500 kg batch runs, and the scale reads 2,420 kg once it's done.

The 80 kg gap comes from somewhere real: evaporation, moisture change, residue left in equipment, spillage, sampling pulled for testing, or ordinary process variation. Whether that gap becomes a financial variance, and how large, depends on valuing it at a standard or expected cost, which the costing chapter below covers. The quantity gap alone doesn't do that automatically.

Process loss, scrap, and rework

These terms describe related but distinct ideas, and industries don't always draw the lines the same way. Expected process loss is planned, built into the yield assumption. Unexpected loss is the gap above and beyond that plan. Scrap is output that can't be used or sold as intended; waste is scrap with no further use at all. Shrinkage is a quantity reduction from process loss, handling, or counting error. Rework is output that missed spec but can be reprocessed into acceptable product rather than discarded.

None of this is a rounding error. The loss assumption feeds material requirements, purchasing quantities, expected cost, and inventory consumption all at once, so getting it wrong distorts all four together.

Unit-of-measure conversions

The vitamin premix is purchased in 25 kg bags, tracked in inventory as kilograms, and consumed by the formula in grams. Finished output is measured in kilograms, but a beverage version of the same formula would be measured in liters. Weight-to-count and weight-to-weight conversions, grams to kilograms, kilograms to bags, are the straightforward mechanics covered in the BOM guide's units-of-measure section. Weight-to-volume is a different, riskier category.

Potency and variable-strength ingredients

Some raw materials have an effective strength that varies from lot to lot even when the labeled material is identical. The vitamin premix scaled to 50 kg above assumes standard potency. The lot that actually arrives tests at 80% potency.

Real potency adjustment can get more involved than this: blending multiple lots of differing strength to hit a target, adjusting for moisture interacting with potency, or working within a regulatory tolerance band. Not every process manufacturer needs it; it matters where a raw material's effective strength genuinely varies lot to lot, and doesn't apply where inputs are consistently specified.

Co-products and by-products

A single run doesn't always produce exactly one output. A primary product is the main intended output. A co-product is an additional output that's economically meaningful and intentionally produced, both are legitimate objectives of the run. A by-product is secondary: it may have value and sometimes gets sold or reused, but isn't the point of the process. Where the line falls between co-product and by-product is usually a matter of economic significance and intent rather than a fixed technical rule, and can shift as a secondary output becomes more or less valuable over time. Multiple outputs from one run mean inventory has to track more than one resulting item, yield has to be measured across all of them, and cost has to be allocated across outputs somehow, an allocation method that depends on the industry and the outputs' relative value rather than one universal formula.

Product and production control

The 2,500 kg batch is now a physical thing sitting in a warehouse, with a number attached to it. This chapter is about what that number has to be able to answer.

Lot tracking and traceability

Process manufacturers need to answer two questions, in opposite directions.

Forward: which finished batches contain material from a specific ingredient lot?

Vitamin Premix Lot VP-2205
        ↓
Production Batch NP-2500
        ↓
Finished Lots FG-2500A and FG-2500B
        ↓
Customer shipments

Backward: which ingredient lots and suppliers fed a specific finished batch?

Customer complaint on FG-2500A
        ↓
Batch NP-2500
        ↓
Ingredient lots consumed (including VP-2205)
        ↓
Supplier sources

Both directions serve the same purpose: scoping a problem accurately. A recall, a quality investigation, or a customer complaint starts from one known point, a lot, a batch, or a complaint, and traceability turns that single point into an accurate list of everything actually connected to it, rather than a guess that's too broad or a connection that gets missed. Specific requirements vary by industry and regulation; the forward-and-backward mechanism is the general concept underneath any of them.

Formula revisions and production history

Formulas change for ordinary business reasons: an ingredient gets substituted, a supplier changes, a cost-reduction effort swaps a component, a regulatory requirement forces a change. The operational question isn't just "what's the current formula?" It's "what formula was actually used to produce this batch, made six months ago?" Knowing only the current version doesn't answer that. Revision control and effective dates, the same mechanism covered for BOMs in the BOM guide, are what make the historical answer recoverable instead of a matter of institutional memory.

Quality holds and release

Batch NP-2500 finishes production and goes on quality hold pending test results before it can ship or even be consumed into another batch.

Specifications, sampling plans, and acceptance criteria vary by industry and product, specific enough that they're outside the scope of a general guide like this one. What matters structurally is that hold status is a real inventory state, not a note attached after the fact.

Shelf life

Both ingredients and finished output carry expiration windows that constrain when they can be used or sold. An ingredient that expires before a scheduled run can consume it is a planning failure, not a quality failure, one that shows up in the next chapter. Finished output has the same constraint working forward: a batch with a six-month shelf life can't sit unsold for eight.

Planning process manufacturing

Demand for 2,500 kg of nutritional powder has to turn into a purchase order for protein, flavoring, and vitamin premix before any of the previous two chapters can happen at all.

Demand and MRP in a process environment

The mechanism is the same one covered in full in the MRP guide: demand creates a production requirement, that requirement explodes through a formula, the same role a BOM plays in discrete manufacturing, into ingredient requirements, those requirements net against on-hand inventory, and the shortfall becomes a planned purchase or production order, offset by lead time. For the 2,500 kg batch, that means MRP resolving the scaled formula (1,750 kg base powder, 500 kg protein, 175 kg flavoring, 50 kg vitamin premix, 25 kg processing aid, plus packaging) against what's already in stock, and generating purchase orders for the rest.

What changes in a process environment is the set of constraints layered on top of that same gross-to-net logic: yield assumptions determine how much input a target output actually requires (2,500 kg of target output needs more than 2,500 kg of input, once expected loss is factored in), shelf life limits how far ahead an ingredient can be purchased, and potency may mean the net requirement isn't fixed until a specific lot's test result comes back. None of that changes the underlying mechanism; it changes what counts as a feasible plan once that mechanism runs.

The "offset by lead time" step above depends on knowing how long a given vendor actually takes to deliver, covered in vendor lead time and reorder points, not a guess entered once and forgotten. And once the vitamin premix and the rest of the scaled formula arrive, matching what was ordered against what was received and billed, covered in three-way matching, is what turns a purchase order into a trustworthy lot cost rather than an assumption.

Batch size and equipment constraints

Process manufacturers usually can't produce an arbitrary quantity on demand. A mixer might hold 2,000 liters, a reactor might have a 500-liter minimum fill, an oven might hold 20 trays per run.

Either combination covers demand, but they're not equivalent: batch size affects ingredient timing, scheduling and labor, how much inventory sits between runs, achievable yield, and cost. A feasible production plan is a different question from a mathematically netted requirement, and equipment constraints are exactly why.

Shelf life in planning

The expiration constraint from the previous chapter feeds directly back into this one. Planning against a formula's theoretical requirement without checking whether an ingredient lot will still be usable by the scheduled production date schedules a run against material that won't legally or practically be available when the plan says it will.

Manufacturing inventory and costing

Batch NP-2500 clears its quality hold. What happens to it next, physically and financially, is this chapter.

From raw materials to finished goods

Raw materials
        ↓
Formula
        ↓
Batch production (WIP)
        ↓
Bulk finished product (2,420 kg)
        ↓
Packaging
        ↓
Finished-goods SKUs: 30-serving tubs, 60-serving tubs, single-serve packets

One bulk batch, three packaged SKUs. That one-to-many relationship between a bulk batch and its eventual packaged SKUs is especially common in process manufacturing, and it's a relationship a simple assembly-oriented inventory model may not capture naturally.

Costing a batch: material, yield, labor, and overhead

What the batch actually cost draws on several inputs together, not any single one of them:

• Raw-material consumption. What the batch actually consumed, including the potency-adjusted 62.5 kg of vitamin premix, compared against what the formula expected.
• Yield and material usage variance. The financial expression of the gap between 2,500 kg expected and 2,420 kg actual, which requires valuing that gap at a standard or expected price, not just noting the quantity difference.
• WIP. Value in process during the run, released once the batch completes, following whatever costing method is in use.
• Labor and overhead. Conversion costs applied to the batch on top of material.
• Co-product and by-product allocation. Where a run produces more than one output, some method of splitting shared cost across them.

Not every process manufacturer uses standard costing, and the exact journal-entry mechanics behind any of the above depend on costing method, company policy, and accounting system. Manufacturing WIP, batch costing, and variance are each substantial enough to deserve their own dedicated treatment; the point here is that the formula and the batch's actual results are the operational inputs those processes run on, not that one universal accounting model sits behind all of it.

Manufacturing COGS

Where a batch's cost feeds an item's standard cost, that cost flows through to COGS on sale. Under other costing methods, COGS derives differently, but it always traces back to the same place: what the batch actually consumed, what it actually yielded, and what labor and overhead were actually applied to it. A formula tells you what a batch should cost. The batch record is what tells you what it did.

How the pieces of process manufacturing connect

Follow the 2,500 kg batch from start to finish and a dependency chain emerges, not a list of disconnected topics:

Demand
        ↓
Material planning
        ↓
Procurement
        ↓
Raw-material lots
        ↓
Formula
        ↓
Batch production
        ↓
Yield and quality
        ↓
Bulk / finished inventory
        ↓
Packaging
        ↓
Fulfillment
        ↓
Costing

None of these steps is individually hard. What's hard is that each one depends on the one before it, correctly, in real time: a change in demand has to reach material requirements before purchasing acts on stale numbers. A specific ingredient lot has to reach the batch record before traceability means anything. Actual consumption and yield, not the formula's theoretical figures, have to reach inventory and cost. Quality-hold status has to reach inventory availability, or a plan believes material is usable that isn't. Production results have to reach costing, or COGS reflects what should have happened instead of what did.

A spreadsheet doesn't usually fail at any single one of those steps. It fails at keeping all of them connected, since nothing forces a change in one tab to propagate correctly into the next. A system built primarily around discrete assembly runs into a narrower version of the same gap: it assumes a fixed component count per unit, not a formula that scales proportionally and yields variably, so yield, potency, and co-product allocation end up bolted on rather than built in.

Process-manufacturing terminology

Frequently asked questions