How the model works — the anatomy of the requirement

Band widths are proportional to pathways per month, so the flow is volume-true from referrals to clock stops. The four boxes on the right change units (appointments, sessions, beds, tests), so those arrows carry the conversion formula instead of a width — the same formulas the levers control. The dashed arrows mark work driven by referrals rather than clock stops: first appointments and diagnostic tests happen at the START of a pathway, so they don't surge with backlog clearance.

The same picture in six sentences

1. Referrals arrive. In …, … pathways join the queue (GP and other referrals, growing each year).
2. Some leave without treatment. About … (…/mo) are discharged after not attending, turn out to be duplicates, are treated elsewhere, or choose not to proceed. This share is calibrated from published data, and it is a lever — it is the single most sensitive assumption in the model.
3. What is left is effective demand. … pathways a month genuinely need a clock stop — treatment, or a clear decision that treatment isn't needed.
4. The backlog adds extra work. To hit the milestone dates the trust must also close … pathways a month from the existing waiting list (the amber band). This clearance work is why required activity runs ABOVE demand until the list reaches its sustainable size — and why the requirement falls back towards demand after April 2029.
5. Clock stops become capacity. Using the trust's own conversion ratios, the selected month's requirement means … outpatient slots, … theatre sessions, … elective beds and … diagnostic tests.
6. Compare with today. The trust currently delivers … clock stops a month; the selected month requires …. The difference (+…) is the capacity ask — the red line on the green bar.

Levers (the same levers as the model pages — the picture updates as you move them)

Where every number comes from

Footnotes — the ideas behind the picture

Three pieces of machinery sit underneath this diagram. Each has its own plain-English explainer:

• Census vs flow — the two cameras. RTT and diagnostics photograph the queue (how old is everyone waiting today?); cancer and A&E score each month's finishers (how fast were they?). One queue, two cameras — this is why the standards read differently.
• New and follow-up attendances — the tandem queue. Why first appointments follow referrals while follow-ups follow treatments, and what that does to clinic templates during a backlog clearance.
• The shape factor k. Real waiting lists aren't textbook-shaped; k measures how far today's list differs from the theoretical shape and scales the sustainable list size accordingly. It is held at today's value by default and exposed as the booking-discipline lever.

The boundary of what the model can support is written down in the Known limitations section of the README — steady-state queueing assumptions, no within-year seasonality, modules don't compete for shared capacity, and workforce counts exclude bank/agency staff.