Proof — Benchmarks

Build a comparison

live against the governed dataset · save it when it earns its place

—

measured, strictly compatible · live denominator

—

median · t/ha

—

middle half of the spread (p25–p75)

Saved benchmark views

context comparisons — never scores or recommendations

Benchmark	Definition	Members	State
BM-UK-WW-YIELD-2526	winter wheat yield · UK · 2025–26 · weighbridge-compatible	25	live
BM-EE-WW-YIELD-2526	winter wheat yield · East of England subset	13	live
BM-UK-WW-GHGHA-2526	winter wheat CO2e/ha · IPCC tier 2 boundary	7	live
BM-UK-WW-SOM-2526	soil organic matter · LOI method	4	suppressed · below k=5
BM-INT-WW-YIELD-2526	international winter wheat yield · methods aligned · 6 countries	25	live
BM-SC-OSR-YIELD-2526	OSR yield · Scotland	0	no matching records — shown honestly

Distribution · BM-EE-WW-YIELD-2526 with your record plotted

Open source case files

n=13 members · winter wheat · East of England · 2025–26 · weighbridge method · confounders: variety, drilling window, soil series · excluded: 12 records (method or basis mismatch — each with its reason) · spread shown, never a league table.

Previous seasons · Long Meadow yield

your own history · corrections visible

t/ha · weighbridge, same method all seasons · 2023 was the dry year — context, not excuse · the 2024 value carries a posted correction, marked, history intact.

The same measure at every scale

click a row to open its records

Your field · Brook Ground 8.1 t/ha · measured

Your farm · 4 wheat fields med 8.0 · n=4

East of England · compatible cohort med 8.0 · n=13

UK national · compatible cohort med 8.0 · n=25

Same metric, same method, honest n at every scale — compatible cohorts only, k=5 throughout. Each row opens its source records.

Programme comparison · authorised cross-client

two advisory approaches · winter wheat · 2025 · within your authority only

Overlapping spreads, denominators visible — no verdict on which approach "wins". Confounders (soil, variety, drilling window) and the excluded records open from here. Visible only inside your authorised scope.

What every benchmark exposes

Membership definition

crop · practice · region · season · method — inspectable, versioned

Denominator

always on screen, live as filters change

Metric / unit / basis / method / period

carried with every value

Included and excluded records

each exclusion carries its exact reason

Confounders and limitations

named beside the number, not in a footnote

Suppression state

k=5, fail-closed, with Broaden-filters action

Custom benchmarks & targets

private by default

Definition	Owner · version	Status
Own-farm 3-season baseline · yield	you · v2	in use
CO2e/ha reduction pathway to 2030	you · v1 · deadline 2030	private
Contract spec · Nunhill low-carbon premium	Nunhill Mill · v3 · mapped	reviewed & shared

One engine everywhere: owner, version, effective dates, metric/unit/basis, method and boundary, denominator, inclusions/exclusions, limitations — target value, pathway and deadline optional. Private until explicitly mapped and reviewed, so an aspiration can't pose as a standard. Progress always shows coverage beside status.

Benchmark rules: comparisons are context, never verdicts. Averages appear only for compatible measures with denominator, unit, period, range, method and source cohort. Every aggregate opens back into the underlying case files — the number is a doorway, not a dead end.