Post 4: Start With One Decision, Not One Platform

70f064a3 7462 49ed bf6b a723c8a5fa6d

The unit of transformation isn’t the machine, the platform, or the model. It is the recurring decision. Here is how to choose your first one.

Every transformation needs a unit: the thing you improve, measure and scale. For most of the last decade, manufacturers chose the wrong one. Some chose the machine (“connect every asset”). Some chose the platform (“implement the MES”). Lately, many have chosen the model (“deploy AI in quality”).

The 99% Factory chooses something smaller and more powerful: the recurring decision.

From the book: Chapter 1 · The Perfect Storm and Chapter 10 · Decision Architecture, in The 99% Factory: From Output to Outcome.

Why decisions, not technology

Every outcome a customer values- quality, delivery, responsiveness, traceability- is the cumulative result of thousands of operational decisions made well or badly. When to change a tool. Which order to run. Whether a batch is fit to release. When to intervene on a drifting process. Whether to accept a supplier’s lot.

Product is the output. Decisions are the mechanism. Once you see decisions as the unit, the only questions that matter become clear:

  • Which decisions drive our outcomes?
  • How good are they today?
  • What would make them better and faster?

Technology then has a job description. It earns its place by improving a named decision, or it does not get bought.

Anatomy of a decision

In the book, I describe every decision as a chain with five links:

Event → Information → Decision → Action → Outcome

A press begins to drift (event). Temperature, force, and cycle time signals show it (information). Someone decides to change the tool (decision). The change happens (action). Scrap falls, and the traceability record is clean (outcome).

Most decisions in an SME break somewhere along this chain. The event goes unseen. The information arrives late or in a form nobody trusts. The decision has no clear owner, or waits on a manager’s sign-off that adds no real judgement. The action does not follow. Mapping the chain for one decision, and marking where it breaks, is often the most valuable hour a leadership team spends all quarter. That is exactly what the Decision Map in the Executive Playbook is for.

What makes a good first decision

Not every decision is a good place to start. From experience, the best first decisions share five characteristics.

  1. It recurs. Daily or weekly, not once a year. Recurring decisions give you enough cycles to measure improvement within ninety days.
  2. It touches a customer outcome. If you improve it, a customer would notice: fewer escapes, faster response, cleaner audits.
  3. It is costly when wrong. Scrap, rework, late shipments, lost accounts. The cost of a bad decision should be visible in money.
  4. It depends on too few people. Ideally, the knowledge behind it is concentrated in one or two heads. Fixing it also reduces a key-person risk.
  5. It needs only a little data. A handful of signals can improve the best first decisions, not a plant-wide data lake.

Decisions that fit all five appear in almost every factory. In stamping, it is often tool change. In food processing, batch release. In injection moulding, defect troubleshooting. In job shops, order sequencing.

Three examples from the book

Meridian Stamping chose when to change a tool. If they changed too early, they wasted tooling. Too late, it shipped defects and failed traceability audits. And it lived in the ear of a retiring setter.

Sunda Foods chose whether to release a batch as safe and traceable. Three systems held three versions of “yield”, and the release decision stalled while people argued.

Halcyon Plastics chose why a part is flashing or short-shooting, and which parameter to change. Only the founder could answer, so they turned away new business.

None of these firms began by buying a platform. Each started by naming a decision, then built just enough connected information, captured knowledge, and technology to improve it. Each became a thin slice through the whole factory, from the customer outcome down to the machine, that could be funded, measured and repeated.

How to choose yours in one meeting

Here is a format I use with leadership teams. It takes about ninety minutes.

Step 1: List ten recurring decisions that most affect your customers. Include who makes each one today.

Step 2: Score each from 1 to 5 on four questions: How often does it recur? How costly is it when wrong? How much does it depend on one or two people? How little data would it need to improve?

Step 3: Pick the top one. Resist the urge to pick three. An SME’s transformation capacity is finite, and one decision done well beats three done badly.

Step 4: Fill in the Outcome Canvas for that decision: the pressure, the customer outcome in their words, the decision, the owner, the evidence that would make it observable, and what “better” looks like in ninety days.

Step 5: Stop one technology discussion. If any system purchase is under discussion that is not tied to a named decision, pause it until it is.

What changes when you do this

Three things tend to happen when a team commits to one decision.

First, the conversation changes. The CFO can fund a capability tied to a measurable outcome. The engineer can design the smallest system that serves it. The operator can see how it makes the shift easier.

Second, vendors become easier to evaluate. “Which of our decisions will this improve, and by how much?” is a question many proposals cannot answer.

Third, momentum builds. A visible improvement in one decision within a quarter does more for the next investment case than any roadmap.

This week’s step

Gather your leadership team and list the ten decisions. You will probably find the first one in plain sight, made every day by someone who’s been doing it for twenty years and has never been asked how.


Your next step: Outcome Canvas