# Delivery percentiles with a small completed sample

Illustrative planning brief; no automatic product import.

Synthetic example: Five completed same-type items took 2, 3, 4, 6 and 10 elapsed days under one clock. A draft promises the next item will finish within its sample percentile.

## Decision

Report the descriptive 80th percentile as six days under nearest-rank and decline to treat it as a calibrated promise.

## Owned work

- Select a consistent sample
  - Owner role: Report owner
  - Acceptance evidence: Five completed same-type records use the same start/finish boundary and date window.
- Show the rank calculation
  - Owner role: Analyst role
  - Acceptance evidence: Sorted values and ceiling(0.8×5)=4 are visible; the fourth value is six.
- Review the forecast claim
  - Owner role: Delivery lead
  - Acceptance evidence: The note identifies small-sample limits, unfinished exclusions and changed work conditions.

## Workflow

1. Keep comparable completed records and disclose exclusions.
2. State a percentile convention, sort values and choose the stated rank.
3. Treat the result as descriptive evidence; review changes before using it in a planning discussion.

## Judgment

A small completed sample omits unfinished work and does not establish an 80% future success probability. Other percentile conventions can give different values.


## Worked reporting note

Historical sample note: n=5; nearest-rank p80=6 days. No calibrated future deadline confidence is asserted; check unfinished work and changed conditions.


## Working artifact

- Observed / Five completed durations under one clock / Supports a descriptive nearest-rank result.
- Unknown / Future-item completion probability / Do not turn this small sample into a promise.
- Next check / Inspect unfinished ages and work-type changes / Keep limitations attached to planning use.


## Workflow questions

### Why is the result six days?

Nearest-rank selects position ceiling(0.8×5)=4 in the sorted list: 2,3,4,6,10.

### Is six days a deadline guarantee?

No. It is a descriptive sample value, not calibrated future probability or accepted scope.

### Should unfinished work be included as finished durations?

No. Report its age separately and disclose the completed-only sample boundary.

## Product connection

Use this manual measurement rule as context when planning or reviewing owned work in TeamBoostAI. Confirm the fields and reports available in your account; the worksheet is not imported automatically.

Confirm account availability before adopting this manual outline.


## Worked measurement study

All example records are synthetic; manual worksheet, no automatic import.

### Inspect the invented records

Completed item | Elapsed days
--- | ---
A | 2
B | 3
C | 4
D | 6
E | 10

### Derivation

Sorted sample [2,3,4,6,10]; n=5; nearest rank for p=80 is ceiling(0.8×5)=4; selected value=6 days. At p=85 rank=5 and value=10: one rank step changes the result substantially.

Historical sample note: n=5; nearest-rank p80=6 days. No calibrated future deadline confidence is asserted; check unfinished work and changed conditions.

Calculator inputs and convention: {"mode": "rank", "values": "2,3,4,6,10", "percentile": 80, "unit": "elapsed days"}

### Evidence boundary

Observed / Five completed durations under one clock / Supports a descriptive nearest-rank result.
Unknown / Future-item completion probability / Do not turn this small sample into a promise.
Next check / Inspect unfinished ages and work-type changes / Keep limitations attached to planning use.

### Method references

Google SRE monitoring: https://sre.google/workbook/monitoring/ — Distribution summaries; no task-delivery forecast is inferred from service latency guidance.
The Kanban Guide, May 2025: https://kanbanguides.org/the-kanban-guide/2025.5/ — Clock and work-item definitions; examples and local rules are our own.