Course map + field guide
Theory of Constraints for service flow
Find the one condition that currently limits more goal achievement, organize the system around it, then look again.
The diagnostic frame
| Question | Service-business translation | IT-Premium example |
|---|---|---|
| What is the system? | Name its boundary and keep materially different flows separate. | Recurring services and RFC/projects are distinct streams sharing capacity and cash collection; incidents and service requests are a shared-capacity demand loop, not a third revenue stream. |
| What is its goal? | Define the result the system exists to create now and sustainably. | Verified customer value, then retention/renewal or commercial closeout—not merely a closed ticket or collected cash. |
| What is the flow unit? | Choose what moves from demand to goal completion. | A retained service period or an approved, delivered, invoiced RFC. Use Абонплата to separate recurring income, then normalize contracts for MRR. |
| What limits more goal units? | State a falsifiable constraint hypothesis, not merely the largest queue. | Hypothesis: clarification-to-approved-plan flow; 17 of 30 RFCs sat in clarification. |
| What protects the goal? | State necessary conditions that an intervention must not damage. | SLA and recurring retention cannot be sacrificed to accelerate project work. |
Map at two levels
| Level | Purpose | Minimum content |
|---|---|---|
| Level 0 — management | Make end-to-end outcome, queues, ownership, and shared capacity visible. | Trigger, customer-facing outcome, exit condition, and feedback loop. |
| Level 1 — improvement | Diagnose the problem area without reducing flow to engineer effort. | Owner, entry/exit criteria, touch time, wait time, WIP, rework, and system/information used. |
The Five Focusing Steps
Identify
Find the current system constraint relative to the goal. Use queue age, blocked time, demand/capacity, and downstream starvation—not queue size alone.
Exploit
Get more goal output from the constraint without major new investment: protect its time, improve input quality, remove avoidable waiting, and use explicit priorities.
Subordinate
Change every other policy, release rule, priority, and metric that conflicts with the exploitation decision. Avoid flooding the constraint with excess WIP.
Elevate
If demand still exceeds effective capacity, add capability, authority, automation, or structural capacity. Elevation usually costs money or creates trade-offs.
Repeat
When the constraint breaks or moves, return to identification. Do not let the old operating rules become an inertia constraint.
Evidence ladder
| Signal | What it says | What it does not prove |
|---|---|---|
| Large queue | Work accumulates at or before this phase. | That the phase itself has insufficient capacity. |
| Old work / blocked days | Flow is persistently delayed. | Whether the cause is internal capacity, policy, or customer dependency. |
| Touch time versus elapsed time | How much delay is actual work versus waiting. | Phase-transition history is still needed to calculate elapsed lead time and wait time. |
| Downstream starvation | Upstream is not feeding enough ready work. | Which upstream rule causes the starvation. |
| Controlled intervention | Changing one mechanism changes throughput while guardrails hold. | That the same constraint will remain forever. |
Course sequence
Weekly experiment template
Constraint hypothesis:
Goal / flow unit:
Exploit decision:
Subordination rules:
Leading measure:
Outcome measure:
Guardrail:
Expected signature if correct:
Disconfirming result:
Review date:Primary method: Goldratt Research Labs, Introduction to TOC. Running case: IT-Premium Value Streams & Flow Constraints, refreshed 19 July 2026.