Planning parameters

From Delft Solutions
Jump to navigation Jump to search

Planning parameters

This article defines the numbers used for long-term capacity planning: planning work beyond the currently in-progress scope. It exists so that roadmap dates are derived from a shared, versioned set of parameters rather than re-argued each time.

Scope: these parameters are for planning at the horizon level — sizing inbound work, projecting when a project can start, and testing whether a proposed commitment fits. They are not a substitute for a project plan once work is scheduled in A-Dato Lynx.

Key numbers

Parameter Value Derivation
Throughput 55 pts/week Story points completed per calendar week, averaged over the trailing six weeks. Includes SRE work.
SRE inbound rate (λSRE) 46 pts/week Median story points of SRE work arriving per week across 2026. Measures demand, not completion.
Project capacity 9 pts/week Throughput minus SRE inbound rate. The capacity actually available for client and internal projects.
Unestimated overhead share 25% of working hours Meetings, administration, invoicing, supplies, and other work carried no story points. Stable across multiple years in Kimai.
Expected size — SRE item 11 pts Historical median story points for a single SRE item. Used to size an unscoped incident.
Expected size — client project 122 pts Historical median story points for a client project. Used to size an unscoped project without producing a project plan.
Expected size — other internal project 56 pts Historical median story points for an internal project.

Using the parameters

Sizing unscoped work

Multiply the count of expected items by the median size for its type. This gives a planning-grade estimate without a project plan. It is valid in aggregate, across several items; it is not a commitment for any single item.

Projecting elapsed time

elapsed weeks ≈ project size ÷ project capacity

At 9 pts/week, a 122-point client project is ≈14 weeks of elapsed time assuming it is the only project in flight. At the 2026 median (5.1 pts/week) it is ≈24 weeks. Two concurrent projects do not halve either figure — they share the same remainder.

Known limitations

These are open issues, not settled parameters. Each affects the reliability of the numbers above.

Velocity is credited to the week work was completed.
Story points are credited to the week in which the estimate was entered, not the week the work was performed. This creates artificial spikes in the rolling average.
Throughput is a level, not a rate
55 pts/week is valid only for the staffing and availability configuration that produced it. It cannot distinguish a slower team from a smaller available week. Normalising to points per available person-day would make time-off and headcount changes arithmetic rather than judgement.