Working model · candidates only

The smallest independently testable report unit.

That is the current working definition of a report “atom.” It is a research lens—not a published schema, API boundary, package, or promise that these candidates will survive evaluation.

Eight candidate atoms.

Each candidate names a job, contract-shaped inputs and output, a principle linkage, and a falsifiable evaluation idea.

Candidate 01

Metric definition

Keep one measure interpretable and reproducible.

Inputs: Formula, grain, population, period, owner
Output: Versioned definition and applicability notes
Evidence linkage: MEASURE-001 · N/H
Evaluation idea: Can an independent reviewer recompute it?

Candidate 02

KPI snapshot

State current performance in a declared comparison frame.

Inputs: Governed value, target, prior, freshness
Output: Value, unit, frame, status
Evidence linkage: VIS-TASK-001 · E1/E2/F
Evaluation idea: Can readers retrieve and classify it accurately?

Candidate 03

Comparison / variance

Make a material difference explicit without distorting scale.

Inputs: Actual, comparator, frame, threshold
Output: Absolute and relative variance
Evidence linkage: VIS-COMPARE-001 · E1
Evaluation idea: Do equivalent frames avoid unjustified reversals?

Candidate 04

Trend / signal

Separate observed movement from evidence of process change.

Inputs: Time series, baseline, assumptions
Output: Trend statement and signal state
Evidence linkage: VARIATION-001 · P/pending
Evaluation idea: Measure false alarms and missed shifts.

Candidate 05

Driver explanation

Connect an outcome to supported contributors without overclaiming cause.

Inputs: Outcome, decomposition, contextual evidence
Output: Observed drivers and bounded inferences
Evidence linkage: NARRATIVE-001 · F/G
Evaluation idea: Can readers separate fact from inference?

Candidate 06

Recommendation / decision

State a proposed choice, rationale, conditions, and owner.

Inputs: Options, evidence, constraints, risks
Output: Decision prompt or recommendation
Evidence linkage: ARGUMENT-001 · F/P/G
Evaluation idea: Can reviewers reconstruct and challenge the case?

Candidate 07

Uncertainty disclosure

Make decision-relevant limits part of the result.

Inputs: Uncertainty type, range, sensitivity
Output: Qualification tied to consequences
Evidence linkage: UNCERTAINTY-001 · E1/E2/F
Evaluation idea: Are confidence and escalation calibrated?

Candidate 08

Provenance note

Let a reviewer reconstruct material claims.

Inputs: Sources, transformations, freshness, revisions
Output: Compact trace and review state
Evidence linkage: PROVENANCE-001 · G/H
Evaluation idea: Can sampled values be traced end to end?

Composition hypothesis

A report is more than a pile of units.

Candidate atoms may compose into a reviewed report only when purpose, sequencing, semantic consistency, argument integrity, accessibility, and governance also work at report level.

No candidate is available to install or certified as best practice.