What we score, and exactly how.
No black box. Rules decide every score; you can see the formula and test it yourself. AI only helps explain the result.
Materiality
Which topics matter most?
For your sector we show the 8–10 most relevant topics. Each gets two scores from 1 to 5:
- Business importance — money, customers, legal exposure.
- Impact importance — effect on people and the planet.
Focus here = business ≥ 4 AND impact ≥ 4
Start values come from sector defaults, then adjust for company size, country and customer type. Every topic is tagged E, S or G.
Quadrant
Focus here
Risk
Which risks need action first?
We list 6–10 risks across physical, transition, legal, market and reputation types. Each has likelihood and impact (1–5), and one recommended action.
Score = Likelihood × Impact (1–25)
- Green
1–6 - Amber
7–14 - Red
15–25
Bands are indicative and are confirmed with our advisory team before launch.
4 × 5
20
red
Pathway
What to do, in what order?
Actions are arranged by time — 30, 90 and 365 days — with a plain estimate of saving (low / medium / high) and effort. Triggers come from your answers: own vehicles add fleet actions; a factory adds pollution-control and safety items.
Compliance items are marked applies to you, probably not or check. For example: CSR under Section 135 turns on turnover, net worth and profit thresholds; a POSH Internal Committee applies from 10 employees.
Evidence readiness
How ready are your documents?
For each buyer question we map the document you hold, and mark it Ready, Partly ready or Gap. Partly means something is missing — a date, a signature, a scope.
Readiness % = (Ready×1 + Partly×0.5 + Gap×0) ÷ questions
Shown per theme: Environment, Labour & Human Rights, Ethics, Sustainable Procurement. We show readiness only — never a promised score or medal.
Labour & Human Rights — tap a status to cycle it
88%
How we keep it honest.
- Rules first, AI for wording only
- Formulas visible and easy to change
- Our own wording — never copied questionnaire text
- Documents confidential, never used to train models