LEARN. PRACTICE. GROW.1. Count work transitions
A queue does not shrink merely because many items were closed. Define the period boundary and transitions: opening, closure and reopening. In a synthetic example, 36 initial requests, 24 new arrivals, 20 closures and four reopenings leave 44 open. Reopened items may keep the same identifier; their work state changes. Avoid double-counting a transition when aggregating tools. If counts do not reconcile, identify transfers, cancellations or cutoff differences before attributing the discrepancy to productivity. The formula is complete only when it represents every transition allowed in the exercise.
2. Inspect volume, age and denominator
A stable total of 50 requests may hide items ageing from five to 20 days. Inspect age distribution, blockers and impact alongside totals. To evaluate an L2 pilot, compare measures with clear denominators: 50 escalations in 200 requests is 25%; 30 in 100 is 30%. The count fell while the rate rose. This does not prove that the pilot caused the change: check comparability, complexity and escalation criteria. Choose investigation based on the service question, such as whether L2 resolves work sustainably, rather than merely sending fewer records to L3.
3. Explicit capacity and priority
For local planning, 40 hours minus 18 reserved for incidents, six for training and four for review leaves 12. A 15-hour request requires resolving a three-hour shortfall through scope, date, suitable capacity or authorized reprioritization. Reservations are not promises of exact consumption, but should not silently disappear. A limit of one urgent item in progress requires assessing a new arrival against the ongoing incident’s impact. Priority can change, but the decision should recognize interrupted work, restart cost and authority. Creating an urgent label in every team does not create extra capacity.
4. Confirm integration outcomes
In this lesson’s fictional system, an API returns 202 when it accepts work and publishes a final state later. Integration closes the request prematurely if it treats the first signal as completion. Define the outcome to observe, how to correlate the later response and who handles failure or missing confirmation. Waiting five seconds does not demonstrate completion. Similarly, speeding one team by two hours can add six waiting hours to the next, increasing the path by four. Measure the outcome and full elapsed time, including returns and waiting, to avoid merely local improvement.
5. A commitment across the service
Two suppliers can meet their initial-response commitments while the service remains unavailable. Map diagnostic, repair and validation dependencies. In this model, 12 diagnostic minutes followed by 18 repair minutes and eight validation minutes total 38, missing a local target of 35. Distinct teams do not remove dependencies. Durations are deterministic only for the exercise; in practice inspect waiting, coverage and variability before promising a deadline. During supplier exit, delivering files does not demonstrate usable transfer: the successor should be able to import, interpret and perform a representative task using the materials.
6. Workshop and summary
Prepare a note for a fictional committee: starting queue 50, arrivals 30, closures 28 and reopenings six. Explain why the final total is 58 and what information is missing to choose an improvement. Add another problem: 14 minutes of diagnosis, 21 of repair and seven of validation against a 40-minute target. The path needs at least two minutes of reduction before adding any buffer. Propose an improvement rehearsal or authorized commitment review while retaining validation. Summary: reconcile transitions, show denominators, respect capacity, check state meanings and measure the service path. Do not turn learning arithmetic into guaranteed production forecasts.
Final queue = initial + new arrivals - closures + reopenings
50 + 30 - 28 + 6 = 58
Prior escalation rate = 50 / 200 = 25%
Later rate = 30 / 100 = 30%
Change = +5 percentage points50 + 30 − 28 + 6 = 58 open requests. Diagnosis 14 + repair 21 + validation 7 = 42 minutes, above the 40-minute target.
Common pitfalls
Comparing counts with different volumes; omitting reopenings; confusing response with restoration; excluding validation to meet a target.
Related topics: Value streams · Supplier management
The service outcome depends on the complete path, including waiting, rework, capacity and validation.
Reference: ITIL 4 CDS public course overview · ITIL 4 CDS; observed syllabus v1.0 mirror, 2025 update comparison pending