Load vs Throughput Lab

Find out where your team’s capacity actually disappears each week.

Load vs Throughput Lab

What this fixes

Everyone is flat out, and nothing is finishing. The obvious conclusion is that you need more people or more effort - but the problem usually isn’t too little work. It’s too little flow.

Activity is easy to see. Flow isn’t. You can see who’s busy. You can’t see the hours of work spent waiting for an approval, sitting behind a dependency, being switched away from, or being redone. Those losses leave no trace - so “work harder” becomes the wrong, obvious answer.

Ask most delivery leads where their team’s capacity disappears each week, and they can’t answer. Not because they’re avoiding the question - but because the system doesn’t show it. This tool does.

How this works

Imagine this: a feature took three weeks to deliver. Everyone remembers people working hard throughout, so the team assumes it simply needed more capacity. Instead, they trace the feature from start to finish. It spent two days in active development - and eleven days waiting for approvals, dependencies and decisions. The work wasn’t slow because people weren’t working. It was slow because the work wasn’t moving.

You follow one real piece of work all the way through your system - a ticket, a feature, a deliverable that recently shipped. Every state it passed through, and how long it actually sat in each.

Then you split its life into two: the time it was <progressing> - someone actively moving it forward - and the time it was <trapped>: waiting (on a dependency, an approval, a decision), being switched away from, or being reworked.

The ratio is the finding. Most teams discover something like 12% progressing, 88% trapped. That number is the point of the exercise - it tells you the effort isn’t missing; it’s stuck.

One honest trace. One system-level answer. You trace a single piece of work to reveal how the whole system behaves. The states it got stuck in are the states everything gets stuck in - that’s why one honest trace beats a dashboard of averages.

Team: one person acts as the facilitator - usually the delivery lead - and runs the trace using the Facilitator Guide while the team supplies what really happened.

Time required

Team ✓   About 20–30 minutes, with a facilitator.

The time required is indicative and will depend on how many states the work item passed through.

The facilitator should read the short facilitation guide first - about 6–8 minutes. It is a downloadable PDF on this page.

When to use

Use it when the team is busy, but delivery is slow - and no one can say where the capacity is going.

Use it whenever:

  • everyone is flat out, but nothing is shipping
  • you’re being asked for more people, and you’re not sure that’s the problem
  • delivery has slowed, and the usual explanations don’t hold up
  • you run regular load reviews and want a diagnostic that names a culprit

What you get

  • The progressing-vs-trapped ratio for one real piece of work
  • A system-level view of where the team’s capacity disappears each week
  • The biggest throughput sink, named - waiting, switching or rework
  • The first thing you’ll change on Monday

You’ve done it well when: the team can point to where its capacity goes each week, name the single biggest sink, and say what changes first.

What you’ll notice

  • How little of the work’s life was spent actually moving forward
  • How much time was spent waiting for a decision - not a technical blocker
  • That adding effort would have fed the same leak

Why this works

Busy people are easy to see, so teams naturally mistake activity for progress. What they can’t see is the time work spends waiting, switching or being reworked. Following one real work item makes those hidden losses visible. One honest trace tells a clearer story than a dashboard of averages, because it shows exactly where throughput disappears.

What you need

One real work item that recently finished. A ticket, a feature, a deliverable - typical, not the worst case. Choose one that finished recently, while everyone still remembers it accurately.

The people who actually touched it. They know where it really sat, and for how long. Without them you’ll reconstruct the ideal workflow, not the real one.

The honest timeline. What actually happened - not what should have happened. Write every state in order, and how long the work sat in each.

A facilitator who keeps it a trace. One person follows the work and holds the discipline: what really happened, not what normally happens. The capture sheet should have space for every state, the time spent there, and whether it was progressing or trapped.

Steps

  1. Pick one real work item (3 min). Team: appoint a facilitator first. Choose something typical that recently finished - not the worst case, not the showcase.
  2. Trace its states (10 min). Follow it from start to finish. Write each state as the team remembers it, in order, before discussing timings. Then add how long it actually sat in each. What really happened - not what normally happens.
  3. Split progressing from trapped (7 min). Label every state as either progressing or trapped. Progressing = someone actively moving it forward. Trapped = waiting (dependency, approval, decision), being switched away from, or being reworked.
  4. Work out the ratio (4 min). Add it up: what percentage of its life was it actually progressing? Write the ratio where everyone can see it, then read it out loud. This is the finding.
  5. Name the sink and the Monday change (6 min). Which single state ate the most time? That’s your biggest throughput sink. Then: “what’s the first thing we’d change on Monday?” Agree one owner for that change.

Output

You leave with:

  • The trace - one real work item, state by state
  • The ratio - progressing vs trapped
  • The biggest sink, named - and the first change you’ll make on Monday

Important rules

  • Trace one real item - and get the ratio. Without the progressing-vs-trapped number, this is reporting, not a diagnosis.
  • Trace what actually happened. Not what should have happened. Teams reconstruct the ideal workflow remarkably fast - keep asking “what really happened?”, never “what normally happens?”
  • Follow the work, don’t log it. This is a trace, not a ledger. You’re following one piece of work on its journey - the states it got stuck in are the states everything gets stuck in.
  • Name a sink, not a dashboard. End with the single biggest source of lost throughput - an explanation with a culprit - and the first thing you’ll change.

Planning to facilitate this activity? Access the resources below to get started.

MIndn
Execution must be smooth. Not chaotic or stressful.
Built by Witmer.