Learning objective
By the end of this lesson, you can distinguish a system constraint from a busy resource and choose exploit and subordinate actions before adding capacity.
Why it matters
Local productivity can rise while completed customer outcomes stay flat. If a non-constraint produces faster, the usual result is more work waiting before the limiting factor.
Core concept: throughput follows the limiting factor
A constraint is the factor whose additional usable capacity would increase achievement of the system goal. Test the claim with end-to-end evidence.
Then apply two low-cost disciplines:
- Exploit: protect existing constraint capacity from incomplete work, interruptions and low-value tasks.
- Subordinate: align release, priority and support elsewhere to the constraint’s needs.
Only then consider elevation through more capacity, redesigned policy or automation.
Visual explanation
Text alternative: preparation can supply ninety items weekly, specialist review can process forty-five and completion is forty-four. Complete inputs and steady release protect the limiting review step.
Worked example
AI triples proposal-draft production, but a security specialist can review only forty-five per month. Thirty per cent arrive incomplete.
Exploit by reserving specialist time for judgement and gating incomplete submissions. Subordinate by releasing proposals steadily and limiting upstream work in progress. Elevate later with evidence assembly for standard controls and cross-training, while the specialist retains approval.
Common mistake
Mistake: identifying the constraint by utilisation alone.
A highly utilised team may be processing avoidable rework caused upstream. Verify whether added effective capacity at that point would increase correct completed outcomes.
Quick check
What is the best response when AI generates more cases for an already overloaded specialist?
A. Release every case immediately.
B. Align release to specialist capacity and improve input completeness.
C. Count drafts as throughput.
D. Maximise utilisation of every other team.
Answer: B. Non-constraint work should protect the constraint and the system goal.
Practical prompt
Name one system goal and one useful throughput unit. Identify the longest persistent queue, then write one exploit action and one rule that subordinates upstream release.
Summary
- Define the goal before naming a constraint.
- Verify that more usable constraint capacity would improve the whole.
- Exploit and subordinate before investing.
- Count correct completed value, not activity.
- Reassess because the constraint can move.
Next lesson
Apply the method in Protect the Security Review Constraint, a proposal workflow where AI drafting has increased work in progress.