2.3 Business Systems, Processes, and Stakeholder Relationships

Processes, systems, and why the distinction matters

A process is a set of activities that transforms inputs into outputs for a customer. A business system is a network of processes whose outputs feed one another, governed by shared resources, shared information, and shared constraints. The distinction is not academic: a Black Belt who treats a process as an island will produce a locally impressive project with no measurable enterprise benefit.

Deming's formulation is the one worth memorizing: a system must have an aim, and the components must be managed to optimize the aim of the system rather than the aims of the components. It follows that the requirements of a sub-process may not align with organizational objectives -- and when they conflict, the sub-process requirement is the one that must give way.

The enterprise process architecture

Most organizations describe their processes in a three-tier hierarchy:

Core processes touch the external customer directly and are where DMAIC projects normally live. Support processes have internal customers and are chronically under-measured, which makes them fertile project ground once the core is stable. Governance processes are where project selection and resourcing decisions are actually made -- which is why a Black Belt who cannot influence them stalls.

Interactive relationships and sub-optimization

Processes interact through four channels, and every one of them is a route by which a local improvement can damage the system:

1. Material and work flow. Speeding an upstream step without matching downstream capacity simply moves the queue and inflates work in process.
2. Shared resources. Two processes drawing on the same equipment, inspectors, or specialists are coupled even when their flows never touch. Loading one starves the other.
3. Information. Forecast accuracy, specification changes, and engineering change orders propagate across process boundaries with lags that create the classic bullwhip amplification.
4. Policy and incentive. A purchasing department measured on unit price buys cheaper material that raises scrap in production. This is the textbook case of a metric creating a defect.

Sub-optimization is the general name for all four failures: improving a component measure while degrading the system measure. The diagnostic question is always the same -- what does this improvement do to the next process downstream, to the processes sharing my resources, and to the measures my counterparts are judged on?

Stakeholders

The exam uses "stakeholder" broadly. A workable classification:

Two implications matter for a Black Belt. First, suppliers are inside the system, not outside it: tightening an internal tolerance without telling the supplier who feeds that characteristic converts your improvement into their defect. Second, regulated processes cannot be improved unilaterally; in medical devices, pharmaceuticals, aerospace, and food safety, a validated process change requires change control, revalidation, and sometimes regulatory notification, and that timeline belongs in the project plan from Define.

Making interdependence visible

Three tools do most of the work:

• SIPOC (Suppliers, Inputs, Process, Outputs, Customers) forces you to name who feeds the process and who consumes its output before you scope anything. It is the standard Define-phase answer to "who else does this touch?"
• Value stream maps extend the view across process boundaries, showing information flow as well as material flow and exposing where queues form between departments.
• Interrelationship digraphs map cause-and-effect links among issues and identify which node is the strongest driver -- useful when the interactions are organizational rather than physical.

Worked example

A distribution center reduces outbound picking time 30% by batching orders into larger waves. Local metrics improve: picks per labour hour rise, and the picking manager's dashboard turns green. System effects: order cycle time rises because early orders in a wave wait for the wave to fill; carrier trailers now load in a compressed window, so dock congestion and detention charges rise; and customer service handles more "where is my order" calls. Net effect on the enterprise measure -- perfect order fulfilment -- is negative.

The failure was not the technique. Wave picking is a legitimate lean tactic. The failure was evaluating it against a component metric without checking the three adjacent processes it touched.

Exam framing

ASQ tags I.A.3 at the Understand level. Expect scenario questions where a described improvement helps one department and harms another, and you must identify the interaction, name the affected stakeholder, or select the systems-level measure that should have governed the decision.