Manufacturing Types and Strategies

The type of manufacturing determines which ERP module you configure, which data objects exist, and which metrics matter. Getting this classification right on day one saves an implementation.

Production types

Discrete manufacturing

Countable, distinct units. Cars, phones, furniture, machinery, garments.

  • BOM — a parts list with quantities
  • Routing — a sequence of operations at work centres
  • Production order — makes N units of one product
  • Assembly-oriented; you can disassemble it back into parts
  • Metric focus: units per hour, first-pass yield, OEE

Process manufacturing

Continuous or batch flows of materials that can’t be un-mixed. Chemicals, food, pharma, paint, cement, refining.

  • Recipe / master recipe / formula instead of a BOM
  • Resources instead of work centres
  • Process order with phases and operations
  • Co-products and by-products — one process yields several outputs
  • Yield and potency — output isn’t a fixed multiple of input
  • Scaling — the recipe scales to batch size, sometimes non-linearly
  • Metric focus: yield %, throughput, batch cycle time, first-time-right

Repetitive manufacturing

High-volume, same product, continuous flow. Consumer goods, electronics, automotive components.

  • No individual production orders — you’d drown in them
  • Production version + rate routing (units per hour, not time per unit)
  • Backflushing — components consumed automatically based on output reported
  • Reporting points — confirm at a few checkpoints, not every operation
  • Metric focus: takt time, line rate, changeover time

Project manufacturing (ETO)

One-of-a-kind, engineered to order. Ships, plants, buildings, custom machinery.

  • WBS / project structure drives everything
  • BOM and routing are created during the project, not before
  • Long lead times, progress billing, percentage-of-completion revenue
  • Engineering change is constant
  • Metric focus: cost to complete, schedule variance, earned value

Lean / Kanban

Pull-based replenishment, minimal formal orders.

  • Kanban containers signal replenishment when emptied
  • No MRP for kanban-controlled items
  • Heijunka (levelling), JIT, one-piece flow
  • Metric focus: WIP, lead time, inventory turns

Planning strategies (the decoupling point) 

                                             Decoupling point
                                                    ↓
MTS   [forecast ──────────────────────────────────] [order] Ship
      Stock finished goods

ATO   [forecast ──────────────────] [order ────────────────] Ship
      Stock components, assemble on order

MTO   [forecast ──────] [order ────────────────────────────] Ship
      Stock raw materials, make on order

ETO                    [order ────────────────────────────] Ship
      Design and make on order
Strategy Lead time to customer Inventory risk Variety possible
MTS Immediate High (finished goods) Low
ATO Days Medium (components) High
MTO Weeks Low (raw materials) Very high
ETO Months None Unlimited

ATO is the sweet spot for configurable products — Dell’s original model. Stock the modules, assemble the combination. It needs a variant configurator and a modular BOM.

Manufacturing execution objects

Object Discrete Process Repetitive Project
Structure BOM Recipe/Formula BOM Project BOM
Process Routing Master recipe Rate routing Network activities
Resource Work centre Resource Production line Work centre / crew
Order Production order Process order Production version + backflush Network / WBS
Confirmation Per operation Per phase Per reporting point Per activity
Costing Order-based Order-based Period-based Project-based

Key manufacturing concepts

Lead time components 

Queue time      waiting before the operation starts     ← usually the biggest
+ Setup time    changeover, tooling
+ Run time      actual processing (qty × time per unit)
+ Wait time     after the operation, before moving
+ Move time     transport to the next work centre
= Throughput time

Queue time typically dominates — often 80–95% of total throughput time in a job shop. This is why reducing batch sizes shortens lead times far more than making machines faster.

Lot sizing on the shop floor

  • Order quantity — how much you make at once
  • Transfer/overlap quantity — you don’t have to finish all 1,000 at operation 10 before starting operation 20. Splitting into transfer batches overlaps operations and can cut throughput time by 50%+
  • Splitting — running one order on parallel machines

Yield and scrap 

Input 1,000 kg → Output 940 kg
  Yield 94%, scrap 6%

Planned scrap:   built into the BOM/routing, expected, in the standard cost
Unplanned scrap: a variance, investigated

Distinguishing planned from unplanned is what makes scrap analysable. If all scrap is unplanned, every order shows a variance and nobody looks at any of them.

Co-products and by-products 

Crude oil refining →  Petrol (co-product)     ← cost allocated
                      Diesel (co-product)     ← cost allocated
                      Bitumen (by-product)    ← credited at NRV, not allocated
                      Waste                   ← disposal cost

Cost allocation between co-products uses relative sales value, physical units, or a fixed apportionment. It’s a genuine accounting decision, not a technical one, and it moves reported product profitability significantly.

OEE — Overall Equipment Effectiveness

The universal manufacturing metric:

OEE = Availability × Performance × Quality

Availability = run time / planned production time      (breakdowns, changeovers)
Performance  = (ideal cycle × units) / run time        (slow running, minor stops)
Quality      = good units / total units                (defects, rework)

Example: 0.90 × 0.85 × 0.98 = 75%

World class is ~85%. Typical is 40–60%. The value of OEE is the breakdown: 75% could be an availability problem or a quality problem, and the fixes are completely different.

The six big losses: breakdowns, setup/adjustment, small stops, reduced speed, startup rejects, production rejects.

Choosing the right module

Ask these questions:

  1. Can you count the units, or do you measure them? Countable → discrete. Measured → process.
  2. Does one process produce multiple valuable outputs? → process manufacturing with co-products.
  3. Do you make the same thing continuously? → repetitive with backflushing.
  4. Is each order unique and engineered? → project manufacturing.
  5. Do you need to trace a batch for recall? → batch management, mandatory in food/pharma/chemicals.
  6. Does the recipe scale non-linearly? → process, with a scaling formula.
  7. Are you assembling from stocked modules? → ATO with a variant configurator.

Most real companies are mixed: a food company does process manufacturing for the product and discrete for the packaging line. Configure both; don’t force one model on everything.

Next