Batch production is a manufacturing mode where a defined quantity of one product, a batch or lot, moves through a sequence of process steps together, and the equipment is then cleaned or changed over to run the next product. It sits in the middle of the production spectrum: more repeatable than one-off job work, more flexible than a continuous line.

Bakeries, breweries, coatings, pharma, food, specialty chemicals, and plenty of machining and packaging operations run this way. Done well, batch production gives you variety without chaos. Done badly, it buries the plant in changeovers, work-in-process, and lot paperwork nobody can trace. This post covers where batch fits on the spectrum, how to size batches, and the traceability discipline that separates well-run batch plants from lucky ones.

Batch vs. Continuous vs. Discrete: What Is the Difference?

The difference is what the product is and how it flows. Discrete manufacturing makes countable units (a valve, a chair) one at a time or on a line. Continuous processing runs an uninterrupted stream (fuel, paper, base chemicals) around the clock. Batch production processes a defined lot together through shared equipment, then resets for the next lot.

BatchContinuousDiscrete
Unit of productionA lot processed togetherAn uninterrupted streamIndividual countable units
Typical productsFood, beverage, pharma, coatings, specialty chemicalsRefining, pulp and paper, base chemicals, utilitiesMachined parts, electronics, furniture, vehicles
ChangeoversFrequent, between every lot or productRare, shutdowns are major eventsVaries, per job or per model run
FlexibilityHigh, many products on shared equipmentLow, the plant is the productMedium to high
Traceability unitThe batch/lot numberTime-slice of the streamSerial number or lot
Core management problemBatch sizing, sequencing, changeoversUptime and process stabilityRouting, scheduling, WIP

Most real plants are hybrids. A food plant runs batch mixing feeding a continuous filling line; a machine shop runs discrete parts in batches to amortize setups. The label matters less than knowing which mode governs each stage, because that decides how you schedule it, a topic the production scheduling hub covers in depth.

The production-mode spectrumWhere batch sits on the production spectrumJOB SHOPone-offs, high skillBATCHlots on shared gearLINE / REPETITIVEpaced discrete flowCONTINUOUSunbroken streamVARIETY: HIGH →→ LOWVOLUME: LOW →→ HIGH
Batch production trades some of the job shop's variety for repeatability, without committing to the volume a line or continuous process demands.

When Does Batch Production Fit?

Batch fits when you make multiple products on shared equipment, demand per product does not justify a dedicated line, and the process naturally works in discrete quantities, a mixer load, a kettle, an oven, a reactor. It also fits when recipes change by product, when regulations require lot integrity, or when shelf life punishes overproduction.

It stops fitting at the extremes. If one product's demand can fill a line year-round, dedicate the line and stop paying for changeovers. If the product is a truly continuous stream, batch thinking just adds artificial start-stops. The expensive mistake is running batch equipment with continuous-flow assumptions, scheduling it with no changeover allowance, then wondering where the capacity went.

How Do You Choose a Batch Size?

Batch size is a tradeoff between changeover cost and everything a big batch costs you downstream. Every changeover consumes capacity, so larger batches spread that cost over more units. But larger batches also mean more inventory sitting longer, slower feedback when something is wrong (a defect discovered at packaging now lives in the whole batch), less scheduling flexibility, and a bigger blast radius if a lot is ever recalled.

Batch-size tradeoff curvesThe batch-size tradeoffBATCH SIZE →COST PER UNIT →CHANGEOVER COST / UNITHOLDING + RISK COSTTOTAL COSTECONOMIC BATCHFaster changeovers push theeconomic batch smaller ←
Changeover cost per unit falls with batch size; holding and risk cost rises. Cut changeover time and the whole optimum shifts toward smaller, more flexible batches.

The most important lever is hiding in the changeover-cost curve: it is not fixed. Cut a 4-hour changeover to 40 minutes with SMED quick changeover methods and the economics that justified monster batches evaporate. Plants that never attack changeover time end up defending big batches as a law of nature, when it is really just the bill for slow setups. Smaller batches then pay a second dividend: fresher feedback for quality, less cash tied up in stock and a schedule that can react when demand moves.

What Is Batch Genealogy and Why Does Traceability Depend on It?

Batch genealogy is the recorded family tree of every lot: which raw material lots went in, which equipment ran it, who operated it, what the process parameters were, and which finished lots came out. When something goes wrong, a supplier flags a bad ingredient lot, a customer complaint arrives, genealogy is what lets you scope the problem to specific lots instead of quarantining a warehouse.

In regulated industries this is not optional. A few anchors worth knowing:

The failure mode is genealogy on paper: batch tickets filled in at end of shift, lot numbers transcribed three times between the mixer log and the ERP. Each retype is a chance to break the chain. This is exactly the class of problem manufacturing traceability programs, and digitizing capture at the station so lot data is recorded once, at the source, are built to fix.

How Do You Run Batch Production Well?

You run batch production well by sizing batches on real changeover economics, sequencing to minimize changeover pain, and recording genealogy at the source. The working sequence:

  1. Define the batch identity. One batch number per lot, assigned at creation, carried on every document, container, and record that touches it. No exceptions, no "we'll write it down later."
  2. Standardize the recipe and the record. Bill of materials, process parameters, and quality checks per batch, in a work-instruction format operators can execute on their worst day.
  3. Measure true changeover time. Last good unit of batch A to first good unit of batch B, including cleaning, verification, and paperwork. This number drives everything downstream.
  4. Size batches on today's changeover time, not history. Revisit lot sizes whenever changeover time improves. If nobody has re-run the math since the last SMED event, batches are probably too big.
  5. Sequence batches intelligently. Light-to-dark colors, allergen-free before allergen-containing, similar setups adjacent. A good sequence can halve total changeover time without touching a single setup step.
  6. Capture genealogy as the batch runs. Material lots scanned in at the point of use, parameters and checks logged at the station, so the batch record is complete when the batch is, not reconstructed two days later.

Where Does Batch Production Go Wrong?

The recurring failure is treating the batch record as after-the-fact paperwork rather than live operational data. When batch status lives on clipboards, scheduling flies blind, quality reviews take days, and a mock recall becomes an archaeology dig. Plants that put batch capture on tablets at the station get the inverse: live lot status, same-shift quality visibility, and batch records that assemble themselves. That is the pattern in the CLS case study paper production logging replaced with real-time capture and automated daily reporting, so the record of what ran is a by-product of running it.

The second failure is letting batch size decisions default to "what we've always run." Changeover time, demand mix, and shelf-life economics all drift. The plants that win in batch mode re-run the tradeoff on a schedule, treat changeover reduction as a capacity project, and keep the genealogy chain unbroken from receiving dock to shipped case.