GGH.
All writing

Flow dynamics · 01

When value lands

Two teams can ship the same number of features and generate very different amounts of user value. Throughput counts what ships; cumulative value measures when each feature reaches users.

By Gavan Grenville-HuntInteractive guideFlow system map

The chart below tracks cumulative value — the running total of feature-weeks users have had access to across everything the team has shipped. The x-axis is elapsed weeks; the y-axis is that running total.

The dashed line shows what the chart would look like if every feature shipped one week after it was built — the fastest timing this team’s throughput allows. The solid green area is what reaches users given the cycle time on the slider. The gap between them is value that exists in the team’s capacity but arrives late: not a throughput shortfall, purely a timing effect.

The slider changes cycle time only. Throughput stays fixed — the team ships one feature per week in both cases.

62%of potential value captured — 38% gap vs. 1-week reference
06121824weeksfeature-weeks
current1-week reference
6 weeks

Throughput is fixed at 1 feature per week — only timing changes.

Try it — three experiments
  1. Set cycle time to 6 weeks
    38% of value left on the table

    The gap between the dashed reference line and the green area is value the team’s capacity could have delivered — but didn’t, because each feature arrived later. The long-run rate of shipping is the same as the 1-week reference; only the timing differs.

  2. Reduce to 3 weeks
    gap falls to 16%

    Same team, same capacity. Each feature arrives 3 weeks sooner. The throughput number is unchanged; the cumulative value captured in 24 weeks is higher.

  3. Reduce to 1 week
    maximum value captured

    The curves converge. All potential value from the team’s capacity is captured. Raising throughput is the remaining lever for generating more.

Features earn from the day they ship

A feature in progress earns nothing. It generates value from the day it reaches users. Every week in development or review delays the value clock by a week.

The chart measures this: for each feature, count the weeks it has been live and sum across all features. That total — feature-weeks in users’ hands — is what users have received from the team’s output. Two teams with the same throughput can have very different totals, because one is shipping each feature weeks earlier than the other.

The timing view

Throughput reports the delivery rate — tickets per week. The timing view tracks when each feature landed and how long users have had it in their hands.

Halving cycle time does not double throughput. Throughput has a ceiling set by capacity — the team can only finish as many features as it has time and skill to build. What changes is when each feature arrives. A team that cuts cycle time from 6 weeks to 2 will report the same throughput numbers before and after; the cumulative value chart tells a different story, value accumulating faster from the point the cycle time fell.

Throughput metrics track the output rate. Cumulative value tracks how much has been received. The two diverge whenever features take weeks to complete.

WIP is value in transit

Every item in progress is a feature whose value clock hasn’t started. A team with 12 features in flight has 12 features sitting between commit and users’ hands. Reducing WIP cuts cycle time, which advances when each feature lands.

Little’s Law describes the relationship: cycle time equals WIP divided by throughput. Cut WIP in half and cycle time halves. At steady state, the same work ships at the same rate — each piece arrives sooner.

The model is ideal; the real world is not. In this chart throughput is fixed regardless of cycle time. In practice, long cycle times tend to erode it. Half-finished code sits long enough to conflict with newer changes, requiring rework. Context fades — a developer returning to a ticket after weeks has to reconstruct where it was and why. Work that has been partially built becomes harder to test and review because the original intent is no longer fresh. The purely mathematical model holds these constant to isolate the timing effect; teams with long cycle times typically see throughput degrade too, which compounds the gap.

Takeaways

  • — Features earn value from the day they ship. Every week in progress delays the value clock by a week.
  • — Throughput measures the delivery rate; cumulative value measures what users have received. The two diverge when cycle times are long.
  • — Cutting cycle time advances when value lands. At the same throughput, every week of earlier delivery is a week of value earned.
  • — WIP is value in transit. Reducing it cuts cycle time and advances delivery. Little’s Law shows exactly how.