The Z-Scale Assessment Series

The Team Throughput Audit (ASTA)

Your system moves at the speed of one team — the bottleneck. This shows whether the rest are feeding it, starving it, or piling up work in front of it — and what fixing that is worth.

Built on Goldratt's Drum-Buffer-RopeiDrum-Buffer-Rope (Goldratt) — how to run a system around its slowest point:
• Drum — the bottleneck; it sets the pace for everyone
• Buffer — a small reserve of work kept in front of the drum so it never starves
• Rope — release new work only as fast as the drum can consume itThe whole system moves at the speed of the drum — so you protect and elevate the drum, not the busy teams around it.
. It's directional, not a stopwatch study — the point is to see reality clearly enough to act. Enter demonstrated rates (what teams actually deliver), not headcount math.

1 · The drum — your bottleneck team

The constraint sets the pace for the whole system. Find it first; everything else is measured against it.

Your demonstrated rate already reflects this, so we don't discount it. It tempers how much freeing up the constraint can realistically elevate output.

Cognitive-load factor: 0.92×

2 · The feeder teams — everyone who supplies the drum

For each team feeding the bottleneck, enter its demonstrated feed rate in the same unit as the drum (bottleneck-ready units per week), and its variability drivers. Those drivers size the buffer.

Feeder team
Feed/wk
Unplanned work
Skill concentration

The Team Throughput Audit · Result

Your system is running at
—
Team by team — is the drum being fed?
The buffer, and what it costs you
Buffer needed
—
days of work held before the drum, sized to your most variable feeder
Added lead time
—
a DORA metric — every buffered day is a day slower to the customer
Flow-efficiency drag iFlow efficiency is the share of a job's total elapsed time actually spent being worked on, versus waiting in a queue.Example: a job that needs 2 days of real work but takes 3 weeks (≈15 working days) to clear = 2 ÷ 15 ≈ 13%. The other 87% is wait, not effort. Typical knowledge work runs 5–15%.
—
the buffer is a queue — pure wait, not work
What fixing it is worth — a year
Total annual throughput you're leaving on the table
—