Skip to content

Methodology

Changelog

We document every change to our scoring model so donors and charities can see what shifted and when.

  1. v3.2 / formula V1.1

    Verified financial refresh and missing-data corrections

    • Joined financial tables by organisation and exact reporting period; preserved explicit zeros and removed estimated balance-sheet inputs from refresh commands.
    • Missing fundraising costs no longer earn zero-cost credit. Unknown filing status receives no filing points. Disclosure coverage is not reweighted and must be considered alongside the score.
    • Income trend points are withheld for gaps, unusually short/long periods and fourfold discontinuities pending comparability review.
    • Historic published regulator reports are linked and dated; their existence alone no longer creates an active-inquiry claim. Confirmed active-inquiry handling remains part of the formula.
    • Held financial profiles refreshed from the complete official CCEW extract, with stable URLs, stored source identifiers and a score-change audit. External independent review of the scoring design remains pending.
  2. v3.1.1

    Star display aligned to score labels

    • Display stars now use the same bands as the labels (90–100 → 5★ Exceptional; 75–89 → 4★ Good; 60–74 → 3★ Needs improvement; 1–59 → 2★ Poor; 0 → not rated) instead of round(score/20), which could show 4★ next to “Needs improvement”.
    • Poor is 2★; there is no 1★ band. Basic profiles are for thin published returns (including many smaller charities and some OSCR/CCNI records), not an exclusion of Scotland or Northern Ireland.
  3. v3.1

    Continuous scoring — removed pass/fail cliffs

    • Every sub-metric now scales continuously between two thresholds instead of a single cutoff (e.g. program-expense ratio scales from 0 pts at 50% spend to full marks at 90%+, rather than jumping to full marks at exactly 75%).
    • Fixes score clustering at 100/100 (was 7.3% of rated charities; now 0.8%) without changing what is measured or the 40/30/20/10 pillar weighting.
    • Trustee oversight and volunteer-to-staff ratio graduated the same way; reserves now taper smoothly on both edges of the 3–24 month ideal band instead of three hard bands.
    • Rationale: a single financial-ratio cutoff is the exact "cliff effect" criticised by BBB Wise Giving Alliance, GuideStar and Charity Navigator in their 2013 joint "Overhead Myth" letter, which called for judging charities on more than a single ratio threshold. We keep financial ratios (Charity Commission data doesn't give us outcomes/impact reporting to weigh instead) but no longer let a fraction of a percentage point flip a score between tiers.
    • All charity records rescored against freshly re-pulled Charity Commission data at the same time (some flagship profiles had been hand-seeded once, pre-pipeline, and never refreshed since — see the data-integrity note on our editorial policy page).
  4. v3.0

    Clarity Score V1 — 100-point pillar weighting

    • Replaced normalized beacon averaging with fixed 100-point model: 40 + 30 + 20 + 10.
    • Statutory inquiry kill switch — automatic 0/100 and regulatory banner.
    • Accountability pillar: filing history, trustee count (≥3), declared policies.
    • Financial Health: reserves 3–24 months ideal, income trend, debt < 30%.
    • Financial Efficiency: program spend ≥75% and fundraising cost thresholds.
    • Community Support: volunteer-to-staff ratio bonus.
    • Star labels: Exceptional, Good, Needs improvement, Poor.
  5. v2.2

    Reserve stability, income trend, and community score

    • Reserve stability hoarding penalty above ~36 months.
    • Added income growth metric and volunteer-to-staff ratio.
    • Fundraising scored as cost ÷ income.
  6. v2.0

    Clarity Score (0–100)

    • Initial four-beacon model from Charity Commission filings.

← Back to methodology