Live Demo

Production Plan Churn

See how much the plan is really moving — and who moved it.

The schedule changed again. Was that a real change, or the system reshuffling itself?

The problem

The production plan changes constantly, but you can’t tell real reschedules from MRP renumbering noise — or whether a planner or the system moved it.

What it does

A week-over-week volatility tracker for the production plan: it separates real reschedules from MRP renumbering noise, attributes each change to planner or system, and pinpoints the materials and weeks taking the most whiplash.

Initializing · Production Plan Churn — real code, loading in your browser

About this tool

Production Plan Churn: what it solves

Every week, MRP regenerates the production plan and thousands of order lines shift. Most of that movement isn’t a real schedule change — it’s renumbering: orders dropping and re-creating with new identities while the actual dates barely move. The few genuine reschedules a planner needs to act on get buried in that churn. Production Plan Churn diffs the plan week-over-week, separates real date moves from renumbering noise, and attributes each change to a planner decision or a system action — so you can see how volatile the plan really is, and where the whiplash concentrates by material and by week. It’s the plan-stability question MRP exception reports were never built to answer.

Most of what looks like a schedule change is the system renaming things

Run MRP and thousands of order lines move. Open the plan on Monday against the plan from the Monday before and almost every row looks different.

Almost none of it is a decision anybody made. Orders drop and re-create with new numbers while the dates underneath barely shift. That is renumbering, and it is the system tidying up after itself.

The handful of genuine reschedules — the ones a planner has to do something about this week — are sitting inside that noise, indistinguishable from it at a glance.

The whole tool is one comparison, run honestly

Take last week’s plan. Take this week’s. Match the lines that are really the same line, then look at what moved.

Matching is the hard part and it is where the answer is won or lost. An order that dropped and came back under a new number is the same work, and a comparison that treats it as a deletion plus an addition reports two changes where there were none. Get that wrong and the tool reports churn that is entirely its own invention.

Every change gets an owner

A date moved. Something moved it. That is either a person making a call or the system reacting to its own inputs, and those two need different responses.

A planner reschedule is a decision, and it stands. A system move is the plan absorbing a change somewhere upstream, and it is worth knowing which upstream.

Splitting them is what turns “the plan is unstable” into something you can act on. Instability caused by planners is a process question. Instability caused by the system is a parameter question, and it lands somewhere else entirely.

Where the whiplash concentrates

Churn is never spread evenly. A few materials take most of it, and a few weeks take most of it, and those are rarely the ones anybody would have guessed.

The tool ranks both. That ranking is the output somebody actually uses, because it turns a site-wide complaint into a short list of materials to look at.

What this page is

A working demo of a tool that is built to order.

It runs on Northpoint Manufacturing, an invented company with invented plants, invented part numbers and invented order data. Nothing here came off any real employer’s system, and nothing here is anyone’s confidential process.

It is one of the factory tools sold to manufacturers on the manufacturer page: built to order for one plant on its own extract, or run once for the written answer. The invented data is what lets anyone open it first. If you run a planning floor and this looks like your Monday, that is the point.

What it will not do

It does not tell you what the plan should be. It reports what changed between two plans you already have. Deciding whether that movement is acceptable is a judgment about your business, your customers and your suppliers.

It sets no stability target. There is no number here saying how much churn is too much, because that number depends on your lead times and your order book and nothing in this data knows either.

It does not reach into a live system. The demo runs on a fixed snapshot in your browser. A real deployment reads exports on a schedule you control.

It reports movement, never a reason. It can tell you a planner moved an order and it cannot tell you why. The why lives with the planner, and a tool that guessed at it would be inventing history.

Built to order for your plant.

This one runs on invented data. Built to order for your plant, on your own extract, and yours outright: the code, the accounts, and a copy that runs with the internet off. Nothing is built before it is bought and there is no license to renew. A factory tool starts at $10,000 per plant. That is a floor per plant for a tool built to order on your own extract, never a quote. What yours costs is agreed before any work begins.

Or run once. The same tool on one extract of yours, and what you get is the written answer it produced rather than the tool. The audits below are that form, and each names the tool behind it.