Decent Mark

I

What we score · Eight pillars, weights fixed

Eight dimensions, and the weights are published before they are used.

The weights sum to 100 and are fixed for the life of rubric v1.x. Every change to the method is published before it takes effect, never after.

  • Weight 12 of 100

    Flexibility & Work Location

  • Weight 18 of 100

    Pay & Reward Integrity

  • Weight 16 of 100

    Job Security & Workforce Practices

  • Weight 12 of 100

    Management Quality

  • Weight 14 of 100

    Workload & Wellbeing

  • Weight 8 of 100

    Hiring Conduct

  • Weight 12 of 100

    Fair Treatment & Voice

  • Weight 8 of 100

    Ethics & Transparency

II

Demerits and merits

Named, numbered rules — not a judgement call.

A demerit deducts points from a specific pillar on a stated evidence requirement for a stated decay period. A merit adds them for a verifiable act. Both are numbered, both are published, and both cite the evidence tier they require.

Severity bands

Minor 2–5, Serious 6–15, Severe 16–30, Disqualifying 31–50. A pillar cannot fall below 0, and merits are capped at +25 per pillar.

Demerit is not flag

A demerit moves the number. A flag is the published string. A company can carry a live Severe demerit with no flag on its page, and that is correct behaviour rather than an omission.

III

What we refuse to score

Some things are not ours to measure.

A scoring instrument is only credible if its edges are published as clearly as its contents.

  • Anything about an identifiable individual, executives included.
  • Access to the courts. A company that defends a claim is exercising a right, and penalising it would make the instrument a punishment for litigation.
  • Sector conditions and forecasts. A forecast is a prediction, and a prediction is not a record of anything a company did. It publishes as Context, which never scores.
  • Anything a company paid us for, because there is nothing a company can pay us for.

IV

The rubric, versioned

Every score names the rubric version that produced it.

66 demerits and 42 merits at v1.2, published in full with their evidence requirements and decay periods. A score card links to the exact version that produced it, so an old screenshot can still be checked against the right method.