Control · C111
Ceiling on initiatives run in parallel: observed throughput, target lead time and range
How many initiatives should a team start this quarter, given the throughput it actually records?
Direct answer
The control linking work in progress, entry rate and time in system holds exactly over the period recorded and assumes no steady state; the ceiling drawn from it does assume the closure rate carries over, so it is published with its interval rather than as a number.
01Set a work-in-progress ceiling from the closure rate and the target lead time, then open a new initiative only when another one closes.
02The rule, the evidence and the next review date stay visible.
How to use this
Move from the question to the deliverable.
How many initiatives should a team start this quarter, given the throughput it actually records?
Set a work-in-progress ceiling from the closure rate and the target lead time, then open a new initiative only when another one closes.
Quarterly work-in-progress record
Relevé et contrôle · SQ-111
Separate what the record proves from what the ceiling bets on
The number of initiatives run in parallel is derived, not settled by conviction. Two things have to be told apart, and practice tends to merge them: the link between work in progress, entry rate and time in system is an exact identity over the period recorded, whereas the ceiling drawn from it rests on the assumption that the closure rate carries over.
Illustrative case: a B2B marketing team of twelve people, an initiative portfolio reviewed each quarter, four quarters on record, and a target lead time brought down to one and a half quarters.
- 01Record
Note by quarter the work in progress at the start, the initiatives started, the initiatives closed and the average work in progress. Requests refused before starting never enter the system, so they never enter the calculation.
- 02Cross-check
Accumulate the area under the work-in-progress curve, then verify that the average work in progress equals the entry rate multiplied by the average time in system. The equality is exact: a gap indicts the records, not the formula.
- 03Bound
Compute the interval of the closure rate from the standard deviation of the four quarters. That interval, and not the caution of the decision maker, sets the width of the ceiling.
- 04Cap
Take the closure rate multiplied by the target lead time, compare current work in progress with both bounds, and open a new initiative only when another one closes.
Work-in-progress record and initiative ceiling
The record tracks work in progress quarter by quarter, derives for the year the identity linking work in progress, entry rate and time in system, then sets the ceiling and the two bounds it sits between.
Average work in progress = area under the curve ÷ duration, entry rate = items entered ÷ duration, time in system = area ÷ items entered, hence average work in progress = entry rate × time in system; ceiling = closure rate × target lead time, the rate being bounded by its mean ± 3.182 × standard deviation ÷ √4| Period | Work in progress at start | Started | Closed | Average work in progress | Cumulative area |
|---|---|---|---|---|---|
| Q1 2025 | 11 | 6 | 5 | 11.5 | 11.5 |
| Q2 2025 | 12 | 7 | 7 | 12.0 | 12.0 |
| Q3 2025 | 12 | 6 | 6 | 12.0 | 12.0 |
| Q4 2025 | 12 | 5 | 6 | 11.5 | 11.5 |
| 2025 | 11 | 24 | 24 | 11.75 | 47.0 |
| Ceiling for a target lead time of 1.5 quarters | — | — | 6.0 ± 1.3 | 9, that is 7 to 11 | — |
01The control is exact, not approximate. Over 2025 the area comes to 47.0 initiative-quarters and 35 items spent time in the system: the entry rate works out at 8.75 per quarter, the time in system at 1.34 quarters, and their product gives back 11.75, the average work in progress on record. No steady-state assumption is needed for that.
02That time in system is not the end-to-end lead time. It is 1.34 quarters over the window while closed initiatives took 1.96: the window cuts off the earlier wait of the eleven initiatives already open on 1 January and that of the eleven still open on 31 December. Mistaking one for the other is the most common error.
03The ceiling, by contrast, is a bet, and it still decides. For a target lead time of 1.5 quarters it comes to 9, between 7 and 11 depending on the interval of the closure rate. Current work in progress, 11.75, exceeds even the upper bound: the decision to cut therefore does not depend on how wide the range is, which is precisely what a range is for.
04Cutting work in progress does not mechanically cut the lead time. Going from 12 to 9 does not bring the lead time down from 1.96 to 1.5 quarters: the identity binds the three quantities without saying which one moves, and work in progress held too tight can lower the closure rate itself for want of supply. The ceiling orders the openings; it promises no lead time.
Illustrative case. The identity is exact over the period recorded and requires no steady state, provided the entry rate counts the initiatives already open at the start of the period. The ceiling, by contrast, assumes the closure rate carries over, which four quarters are not enough to establish. The thirteen requests refused before starting are excluded: counted in, they break the formula.
Reference sources
Coherence controls
Minimum conditions before validation.
The status, the evidence and the owner must be available for review.
The status, the evidence and the owner must be available for review.
The status, the evidence and the owner must be available for review.
The status, the evidence and the owner must be available for review.
Usable deliverable
Quarterly work-in-progress record
Explore the decision further with an interactive control, or start from a structured template to produce your deliverable.
