A guide to the Product Complexity Model

One new item is never one new thing

The model draws a make-and-ship operation as a flow chart and lets you grow it. It exists to show people outside operations what a new product actually asks of the plant.

Open the model

What you are looking at

The chart reads left to right, and it is the life of a product: order the parts, receive them, store them, make the product, package it, inspect it, store it again, pick it, pack it, ship it. The steps never change. What changes is how many different things have to pass through them.

Every thin line is one item that somebody has to buy, make, count, find or ship. Every triangle is one item sitting in stock.

Blue linesFinished items on their way through production, storage, picking, packing and out the door.

Amber linesPackaging. A thin one belongs to a single finished item. A thick one is shared by every item of that size.

Grey linesBuying. One line per vendor into receiving, one line per thing bought into storage.

Triangles and the D shapeStock. Triangles are items in storage. The D is half-made product waiting between steps.

The chart at Day one: a single line running through purchase orders, receiving, storage, one processing step, one packaging machine, inspection, storage, picking, packing and one carrier. — Day one. One offering, one size, one machine, one channel. Seven things to keep in stock and four people. One person can hold all of it in their head.

The chart at Sprawling: hundreds of lines fanning out from storage through four packaging machine types into a second storage pile, then through picking and three kinds of packing to five carriers. — Sprawling. Twenty offerings on the same ten steps. Nobody added a step. The lines are what twenty offerings, four machine types, three package types and five channels multiply out to.

The same plant, counted

The numbers under the chart count what the picture shows.

What has to be managed	Day one	Early	Mature	Sprawling
Offerings	1	8	12	20
Finished items	1	48	120	420
Things to buy	6	93	177	525
Vendors	3	8	15	38
Purchase orders a month	2	4	8	19
Items to keep in inventory	7	141	297	945
Pick faces in the warehouse	1	96	360	1,260
Ways out the door	1	4	6	12
People to run it (estimate)	4	13	26	62
OTIF, on time and in full (estimate)	98%	96%	94%	77%

How to use it

Drag the slider. It runs through four saved states: Day one, Early, Mature and Sprawling. Between them the model fills in the steps, so you watch machines, package types and channels arrive one at a time.

Add or take away one. Under the slider are minus and plus buttons for an offering, a size, a package type and a channel. Press one and the chart redraws, and the numbers under it show what came with it.

See what each click changed. After a plus or minus, a line beside the buttons lists the change: people, OTIF points, cost index, finished items, ways out and rework loops. Red is worse, green is better.

Rework and scrap. Tick it to see what complexity costs in things done twice or thrown away. Red curls with broad arrows are items that went round a step again. At Receive they are returns toward the vendor. Scrap runs along one red line to a rolling bin, then to dumpsters, up to six, as it grows. Both boxes are ticked on load. Untick Rework and scrap to see the plant without it. Animate flow makes the lines crawl in the direction items move.

The team. The small figures at the lower left are the people estimate, one per person. Hard hats are leads and maintenance, caps are operators and pickers, a figure carrying a box ships, a tie is purchasing and a bun is planning. The mix is an approximation from the staffing model.

Change the assumptions. The rollups at the bottom hold every setting: how many ingredients, how many machines, which channels ship which way. Change them and save your own states.

What one more offering brings at Mature: 10 new finished items, 12 new things to buy, 30 new pick faces and 13 more changeovers in every production cycle. That is the conversation the model is for.

What the numbers mean

Finished items: Every distinct thing that can be sold: each offering in each package size.
Things to buy: Every distinct ingredient, packaging component and case that has to be purchased.
Items to keep in inventory: Things to buy plus finished items. Each one has to be counted, stored and kept from running out.
Pick faces: Places in the warehouse a picker pulls from. One per finished item for each unit it ships in: each, case, pallet.
Changeovers per cycle: Stops to switch a machine from one item to the next, counted over one round of making everything once.
Ways out the door: Every pairing of a sales channel with a way it ships. Each has its own routine and cut-off time.
People to run it: An estimate built from machines, offerings, pick faces and the other counts. It is a guide to scale, not a staffing plan.
OTIF: On Time, In Full: the share of orders that leave on time and complete. It falls as complexity compounds, and falls further when Rework and scrap is ticked. An estimate, not a measurement.
Rework % and scrap %: The chance that an item is hit by rework, or scrapped, at some step on its way through. Each stage loses more as it gets more complex, and the losses compound.
Logistics cost index: A relative number, not dollars. Day one is 1. It rises with vendors, stock, pick faces and ways out the door.

What it does not claim

It does not know volume. The model counts how many different things there are, not how much of each gets made or sold. A plant with half the items and the same sales is a different business from a plant with half the sales.

It counts high on purpose. It assumes every offering comes in every size on every machine that can run it. A real plant skips some of those combinations. The shape is right. The totals are a ceiling.

Its rework rates are assumptions. The starting rates, how fast they grow and the OTIF effects are set so the picture looks plausible. They are editable under Rework and scrap rates, and none of them comes from a real plant.

It is generic. There is no company, product or customer in it. The states are built to look like a mid-size plant, and anyone who runs one should recognize their own.

It is not an argument against new products. It is a way to see what a new product costs the operation before saying yes to it.

Guide to the Product Complexity Model v1.3 · built 2026-10-04