Skip to content

The X Standard

What is measured, and what is refused.

A public shape for the method, without a rubric to game and without a certification on offer.

Proposed · not approved

The X Standard v0.1 is proposed and under review. Its thresholds are explicitly not calibrated, its release tests have not been run, and its certification language is unresolved. Nothing on this page has been approved for use on a customer report.

Status as at 2026-09-11 · Sources: X Standard v0.1 (proposed) · xScore contract

The dimensions

What a report is read against.

Seven things, named without weights, thresholds or a formula — those are not settled and are not published here.

  • 01

    Audience and mandate

    Who the report is actually for, and what it was commissioned to do.

  • 02

    The decision

    The specific decision the report exists to support, and the person accountable for it.

  • 03

    The first five seconds

    What a first-time reader sees before they read anything — the leading question, or the absence of one.

  • 04

    The truth of what is shown

    Whether the numbers on screen match what the underlying model and source actually support.

  • 05

    The path through it

    Whether the report leads somewhere a person can act, or stops at the numbers.

  • 06

    The craft

    The ordinary discipline of a well-built report — clarity, consistency, restraint.

  • 07

    Accessibility

    A floor that gates a result rather than a dimension that trades against the others — it never moves the band, it can only stop one.

The evidence ladder

Evidence decides what may be said.

Six kinds of evidence, ranked — no status column, because the order itself is the point.

  1. 00

    Screenshot only

    What the report looks like, with nothing behind it confirmed.

  2. 01

    + the answers of the person who owns the decision

    The decision, the audience and the mandate, taken directly from the accountable person.

  3. 02

    + the report's own definition files

    The measures and their written definitions, as the report itself declares them.

  4. 03

    + the semantic model

    The relationships, grain and keys underneath the definitions, not just the definitions in isolation.

  5. 04

    + how it behaves when run

    The report exercised across real states and filters, not just inspected at rest.

  6. 05

    + evidence of how it is actually used

    Observed use by the people it was built for, not a claim about how it is intended to be used.

A result can never claim more than the evidence that produced it.

The gates

What stops a result

  • 01

    No decision contract means no score at all

    Without an accountable decision on file, nothing is issued — not a low score, no score.

  • 02

    A critical finding suppresses the composite

    While a critical finding is open, the composite is withheld, not shown with a caveat.

  • 03

    An unassessed dimension is left unassessed

    A dimension without enough evidence is marked as such, never interpolated from the others.

  • 04

    A result never outruns its tier

    The evidence tier behind a result caps what the result is allowed to claim, regardless of how the numbers would otherwise read.

The boundary

What a result does not certify.

The boundary

A result does not certify correctness, and it does not certify accessibility conformance.

It says nothing about states that were not examined. The certification language itself is unresolved, and no certification is issued today.

Next

See the standard applied.

The rubric only means something against a real report.

Start with the question

What should your report help you decide?

Two answers are enough to begin. The brief is written in your browser: email it from your own mail app, or save a copy. This site sends nothing on its own.

Goes to mihai@playwithdata.io · No account · No tracking · No network request from this site