Manufacturing & wastage

How to reduce wastage in a food factory (batch-wise system that actually holds)

Updated 8 August 202612 min readnaffo.tech manufacturing teamReviewed by naffo.tech implementation desk

How do you reduce wastage in a food factory?

Reduce wastage in a food factory by measuring it before trying to cut it. Reconcile input against output for every batch, split the gap into named categories (process loss, by-product, rework, rejection, spillage, expiry, packaging), set a tolerance limit per category, and price every loss in rupees. Most factories find 60–80% of their wastage sits in three or four repeatable causes that fixed recipes, FEFO issue and yield-by-shift reporting remove within a quarter.

Key takeaways

"Wastage" is not one number. Until you split it into raw-material, process, rework, rejection, expiry and packaging loss, you cannot assign an owner or a fix.
Batch-level input–output reconciliation is the single highest-return control. If 1,000 kg goes in and 920 kg of sellable output comes out, the 80 kg gap must be named, not averaged away.
Price loss in rupees, not kilos. A 0.4% loss on packaging film can cost more than a 3% loss on whey.
Set a tolerance per loss type (for example process loss ≤ 2%, rejection ≤ 0.5%, packaging ≤ 1%) and alert on the batch that breaches it, not on the monthly average.
Compare yield across operator, shift, machine and season. Variance between shifts is usually a training or calibration problem, not a material problem.
By-products are inventory, not garbage. Whey, buttermilk, cream, trimmings and broken product either have a resale value or a secondary recipe.

Key figures

60–80%: Share of factory wastage typically concentrated in 3–4 repeatable causesBasis: Pattern observed across naffo.tech dairy and food plant rollouts once losses are categorised per batch rather than reported as a single monthly figure.
1–4% of input: Typical unexplained variance before batch reconciliation is enforcedBasis: Gap between issued material and accounted output in plants that record consumption at month-end instead of per batch.

Start by admitting you do not know where the loss is

Almost every food factory that says "our wastage is about 3%" is quoting a plug figure — the difference between what stock said and what the count found. That number cannot be reduced because nobody owns it. The plant head cannot fix "3%". They can fix "paneer batch 214 lost 11 kg because the press cycle ran four minutes short on B shift".

So the first move is not a cost-cutting drive. It is measurement discipline: every batch must close its own material balance. Once that is in place, the wastage number stops being a mystery and becomes a list of named, ownable causes.

The seven places wastage actually hides

Loss is not one event in production. It accumulates across the chain, and each point has a different owner and a different fix. Split your reporting the same way:

Loss type, where it originates, and who owns the fix
Loss type	Where it happens	Typical root cause	Owner
Raw-material loss	Receiving, weighing, storage	Short receipt, moisture variance, no incoming QC, poor stacking	Purchase / stores
Process loss	Cooking, separation, evaporation, cutting	Recipe drift, wrong temperature or time, uncalibrated equipment	Production
Rework	Post-process, pre-packing	Out-of-spec texture, fat or moisture; correctable batch	Production / QC
QC rejection	In-process and finished-goods QC	Contamination, spec failure, foreign matter	QC
Overproduction and expiry	Finished-goods store, distribution	Plan built on hunch instead of orders and shelf life	Planning / sales
Packaging loss	Filling, sealing, labelling, cartoning	Film wastage on changeover, seal failure, label misprint	Packing
Spillage and handling	Transfers, pumping, decanting	Leaks, overflow, untrained handling, no drip recovery	Production / maintenance

A category list this specific does one thing that matters: it makes the reason field a dropdown instead of a free-text box. Free text is where accountability goes to die.

Reconcile input to output for every single batch

This is the control that pays for everything else. At batch close, the equation must balance before the batch can be marked complete.

Batch material balance

Issued input = Good output + By-product + Rework + Rejected + Measured loss + Unexplained variance

The last term should trend to near zero. A persistent unexplained variance means a weighing, recording or issue-control problem, not a production problem.

Two implementation notes that decide whether this survives contact with the shop floor. First, the operator must be able to close a batch in under two minutes on a phone or tablet — if it takes ten, they will batch-enter fiction at shift end. Second, quantities must convert automatically between units (litres, kg, fat-corrected equivalents) or the balance will never tie and the team will stop trusting it.

The formulas to standardise before anyone argues about numbers

Half the wastage arguments in a plant are definitional: production counts loss on input, accounts counts it on cost, sales counts it on finished goods. Fix the definitions once, in writing.

Wastage %

Wastage % = Actual wastage quantity ÷ Total input quantity × 100

Actual wastage excludes recoverable by-product. Counting whey as wastage in a paneer plant will make your numbers look terrible and your decisions worse.

Yield %

Yield % = Good sellable output ÷ Total input × 100

Always state the basis — output kg per input litre, or per input kg. A yield figure without a basis is not a figure.

Wastage cost

Wastage cost = Wasted quantity × (Material cost + Value added up to loss point)

Loss after cooking and packing costs far more than the same loss at intake, because you have already spent energy, labour and film on it. Costing loss at raw-material rate systematically understates the damage.

By-product recovery %

By-product recovery % = Recovered by-product ÷ Theoretical by-product × 100

If theoretical whey is 780 L and you tanked 760 L, you recovered 97.4% — the missing 20 L went to drain and is real money in a plant running 30 batches a month.

Set a tolerance per loss type, then alert on the breaching batch

A wastage report nobody has to answer for is a newsletter. Tolerances turn it into a control. Start with limits you know you can hold, then tighten quarterly.

Illustrative starting tolerances — set your own from your last 90 days of batch data, not from a benchmark article
Loss category	Starting tolerance	Alert when	First thing to check
Process loss	≤ 2% of input	Any batch > 2%	Temperature log, cycle time, recipe version used
QC rejection	≤ 0.5% of output	Any batch > 0.5%	Reject reason code, upstream CCP, incoming material lot
Packaging loss	≤ 1% of packaging issued	Any run > 1%	Changeover count, sealing temperature, film reel lot
Spillage	≤ 0.3% of input	Any batch > 0.3%	Pump seals, transfer lines, drip trays, decanting method
Expiry write-off	≤ 0.2% of finished-goods value	Any SKU-month > 0.2%	Production plan vs orders, FEFO adherence, slow-mover list
Unexplained variance	≤ 0.5% of input	Any batch > 0.5%	Weighing calibration, issue control, data entry timing

Twelve controls, in the order they pay off

01Batch-wise input vs output reconciliation
Non-negotiable, and the prerequisite for everything below. No batch closes without a balanced material equation.
02Standardised recipes and BOMs, versioned
Fixed quantities, temperatures, timings and sequence. Version them so you can attribute a yield change to a recipe change rather than to folklore.
03Yield measured by batch, shift, operator and machine
The comparison is the insight. Paneer at 18% on A shift and 15% on B shift is a three-percentage-point training or calibration gap you can close this month.
04FEFO issue for perishables
First Expiry, First Out — not FIFO. For dairy, cultures, additives and short-shelf-life ingredients, receipt date is irrelevant; expiry date decides issue order.
05Production planned against real demand
Pending orders, actual offtake, current stock, remaining shelf life, seasonality and festival demand. Overproduction is the most expensive loss because you pay full conversion cost and then throw it away.
06Controlled material issue
Issue the planned quantity only. Extra issue is a separate reason-coded transaction. This one change surfaces over-consumption that stock reconciliation hides for weeks.
07Reason-coded rejections
QC failure, machine issue, wrong temperature, contamination, packaging damage, operator error, expiry, spillage. Dropdown, mandatory, reportable.
08Packaging loss tracked per run
Pouches, bottles, caps, labels, cartons, film metres and sealing failures counted against the production run. Packaging is the loss most often invisible in ERP and most visible in cost.
09By-products treated as stock
Whey, buttermilk, cream, trimmings, broken product and off-cuts get an item code, a rate and either a sale channel or a secondary recipe. Anything that goes to drain unmeasured will grow.
10Storage conditions monitored, not assumed
Temperature, humidity, pest control, stacking pattern, cold-chain excursions and warehouse hygiene. One failed chiller overnight can exceed a year of process-loss savings.
11Preventive maintenance on loss-causing assets
Calibration drift, seal leakage, filler over-dosing and cutter misalignment create continuous small losses that never trigger a breakdown — and therefore never get scheduled. Track give-away on fillers specifically.
12Wastage tolerances with per-batch alerts
The feedback loop. Without it, the first eleven controls generate data that nobody acts on.

The Wastage & Yield dashboard to run it from

One screen, one line per batch, left to right along the material flow. If a plant head cannot read this in thirty seconds, it is the wrong dashboard.

Dashboard column order

Raw material input → Good output → By-product → Rework → Rejected → Process loss → Wastage % → ₹ loss

Add filters for date, product, shift, operator, machine and recipe version. The filters are what turn a report into a diagnosis.

Two supporting views earn their place next to it. A yield variance table (product × shift × operator, worst-first) tells you where to send the supervisor tomorrow. A loss-by-category rupee chart (rolling 30 days) tells you whether your biggest problem is actually procurement, storage, production, QC, packaging or finished-goods expiry — which is usually not where the plant assumed it was.

Wire it to the transaction chain or it will rot

Wastage tracking maintained in a parallel spreadsheet decays within two months, because it duplicates entry and disagrees with stock. The loss record has to be a by-product of the transaction the operator was already making. In practice that means a single chain:

Traceable chain

BOM / recipe → material issue → production batch → in-process QC → good output + by-product + rework + rejection → finished-goods QC → batch stock with expiry → dispatch → sale

Every loss is then attributable to a batch, a lot, a shift and a rupee value automatically, and a recall or a customer complaint resolves in minutes rather than days.

This is exactly the chain naffo.tech's manufacturing module implements: recipes with versioned BOMs, batch-wise material issue, by-product allocation that reduces the effective raw-material cost of the main product, QC capture at in-process and finished-goods stages, and batch stock carrying manufacturing and expiry dates through to dispatch. The wastage number is then not a separate report — it falls out of the batch close.

If you are still choosing a system, the demo script in the ideal ERP workflow for a food manufacturing business is the fastest way to find out whether a vendor can actually hold this chain together.

A realistic 30-day rollout

Days 1–3. Freeze the loss category list and the four formulas above. Write them on one page and get the plant head, QC head and accountant to sign the same page.
Days 4–10. Enter recipes and BOMs for your top five SKUs by volume. Not all fifty. Five.
Days 11–20. Run batch close with material balance on those five SKUs only. Expect the unexplained variance to be embarrassing in week one — that is the point, and it is the number that falls fastest.
Days 21–25. Set tolerances from your own first two weeks of data, and switch on per-batch alerts to a named person.
Days 26–30. Add the rupee costing layer and hold the first weekly yield variance review. Fifteen minutes, worst batches first, root cause recorded against the batch.

Then extend SKU by SKU. The plants that fail at this are the ones that try to instrument every product on day one; the plants that succeed prove the loop on five SKUs and let the shop floor see a real number change.

Set up batch-wise wastage control in a food factory

01Freeze your loss categories
Agree a fixed list of loss reasons — process loss, by-product, rework, QC rejection, spillage, contamination, machine fault, packaging damage, expiry, unexplained variance. Nothing gets booked as plain 'wastage'.
02Standardise every recipe and BOM
Lock quantity, temperature, time and sequence for each product version. Operators execute a version; they do not improvise. Version the recipe so yield can be compared before and after a change.
03Issue only the planned quantity
Material issue against the batch equals the BOM quantity. Any extra issue is a separate, reason-coded request, which makes over-consumption visible the same day instead of at stock count.
04Reconcile input to output at batch close
Good output + by-product + rework + rejection + measured loss must equal issued input. Force the operator to close the equation before the batch can be marked complete.
05Cost the loss
Multiply each loss quantity by its material or landed production cost, so the daily report is in rupees. This is what makes the plant head act.
06Set tolerances and alert per batch
Define a limit per loss category per product. Raise the alert on the individual breaching batch while the material, machine and operator are still identifiable.
07Review yield by shift, operator and machine weekly
Fifteen minutes a week on the yield variance table. Investigate the outlier, not the average. Record the root cause against the batch so the same reason cannot recur anonymously.

Frequently asked questions

What is an acceptable wastage percentage in a food factory?

There is no universal figure — it depends on the product, process and how you define wastage. What matters more is a tolerance per loss category set from your own last 90 days of batch data, and an alert on the batch that breaches it. As a starting frame, many plants begin with process loss ≤ 2% of input, QC rejection ≤ 0.5% of output, packaging loss ≤ 1% of packaging issued and unexplained variance ≤ 0.5% of input, then tighten quarterly.

Should by-products be counted as wastage?

No. Recoverable by-products such as whey, buttermilk, cream, trimmings or broken product are inventory with either a resale value or a secondary recipe. Counting them as wastage inflates your loss figure and hides the real problem. Track them as by-product output, measure recovery against theoretical yield, and allocate their value back so the main product's effective cost drops.

What is the difference between FIFO and FEFO, and which should a food factory use?

FIFO issues the oldest received stock first; FEFO issues the stock with the earliest expiry first. For perishable ingredients and finished goods, FEFO is correct, because a lot received later can expire earlier — different suppliers, different remaining shelf life. Using FIFO on short-shelf-life material is a common and expensive cause of expiry write-offs.

How do you find out whether wastage is a procurement, production or planning problem?

Categorise every loss at the point it occurs and chart the rupee value by category over a rolling 30 days. If most of the cost sits in raw-material and storage loss, it is procurement and warehousing. If it sits in process loss and rejection, it is production and QC. If it sits in expiry and finished-goods write-offs, it is planning and sales. Plants routinely discover their assumed culprit is third on the list.

How much detail should an operator enter at batch close?

Good output, by-product, rework, rejection with a reason code, and measured loss — nothing more, and it must take under two minutes on a phone or tablet. If batch close takes ten minutes, operators will backfill it at shift end from memory and your data becomes fiction. Depth comes from consistency across every batch, not from long forms on some of them.

Can this be done in Excel instead of an ERP?

You can prove the loop in Excel for a handful of SKUs, and that is a reasonable first month. It breaks when the loss record has to agree with stock, costing and GST — a parallel sheet drifts within weeks because entry is duplicated. The durable version records loss as a by-product of the material issue and batch close the operator was already doing.

This page also answers

What is an acceptable wastage percentage in food manufacturing?
How do you calculate process loss in a production batch?
How do you track yield by shift and operator?
What should a wastage and yield dashboard contain?
How do you reduce packaging material loss on a filling line?

References

Run this inside one system

naffo.tech is an all-in-one business management and manufacturing platform for Indian SMEs: versioned recipes and BOMs, batch-wise material issue under FEFO, yield, by-product and reason-coded wastage capture, in-process and finished-goods QC, batch traceability to dispatch, GST invoicing with e-invoice and e-way bill, and double-entry accounting that closes live.

Book a demo See the modules