The large figure is the measurement. The chip beside it is that measurement's score out of 10, coloured by band — the number is always printed, so colour never carries the meaning alone. A faded chip marks an illustrative score.
12.4% in italics — a sample value, not a real TeachHub read. Hover any cell for its exact feasibility status from the metric menu.
Role colour marks the band only; the group row beneath it names the sheet's own metric groups within that role. Status colour lives only in the chips.
Scored at the group level, not the raw column level — a role with eighteen columns (Class Head) doesn't get eighteen votes just because Class Head has the deepest metric list. Each of the sheet's own groups (Onboarding, Retention, Setup, Delivery, Assessment, Class quality, Data hygiene, Escalation) contributes one 0–10 score, averaged from whichever of its columns are actually scoreable. Two groups — Recruitment and Community — have no scoreable "For Sure" column yet and sit out of the total entirely.
That leaves Class Head carrying roughly half the composite (4 groups of 8), Volunteer Lead a quarter, D&D Head and the escalation signal an eighth each. The bar under each total shows that split; sort by a role's own column when you want that person's picture.
Only four columns are real: Roster (active count), L1 complete, Dropout, and Turnout — pulled from the same Centre Health Score sheet as before. Turnout is dead (reads 0.00% everywhere) and stays excluded from scoring for that reason.
Everything else in this table is illustrative. Thirty-one columns come from "For Sure" rows in the metric menu that have no live source yet — most are Needs a form or Empty — zero rows in the sheet's own Feasibility column. Their values here are seeded per-centre samples, not real reads; they exist so the shape and density of the table can be judged before any of them get built.
Sign-off rate carries a known trap — a centre that files nothing reads 100% by design upstream — so it's shown but excluded from scoring. Two-week-red escalation is circular by nature: it needs this exact weekly scoring loop to exist for two consecutive weeks before it can be real, so it's shown here as a same-week proxy only.
Four centres break the dropout scale. Above 40% falls off the source sheet's bands and returned #N/A; those centres score 0 there and carry a ▲.