Updated: 30 July 2026
Method should match the question. A news report, process guide, product comparison and original analysis do not require identical evidence, but each should make its basis and material limitations understandable.
1. Frame the decision
Research begins with a defined audience, operating context and decision. The question should be narrow enough to test, and the article should explain what is inside and outside scope.
2. Build the evidence set
We seek the strongest reasonably available evidence. The usual order is primary legal or regulatory material, official records and filings, original data, standards, technical documentation and direct interviews, followed by reputable secondary analysis for context or corroboration. Source selection considers authority, proximity to the fact, date, method, incentives and whether another source can verify the point.
3. Evaluate sources and claims
A source is not treated as conclusive merely because it is official or widely cited. We examine definitions, sample, time period, exclusions, conflicts and whether the source actually supports the claim being made. Contradictory credible evidence is assessed rather than omitted for convenience.
4. Handle data and calculations
Material figures should identify the unit, period, population, currency and source. Transformations such as inflation adjustments, currency conversion, aggregation or percentage change should be reproducible from the stated method. Rounding should not imply false precision. Where practical, a second check is performed on important calculations.
5. Comparisons, scenarios and inference
Comparisons use consistent criteria that are relevant to the stated decision. Inclusion, weighting and scoring rules should be explained when they affect the result. Estimates, forecasts and scenarios identify key assumptions and, where useful, a reasonable sensitivity range. An inference is labelled as analysis and is not presented as a reported fact.
6. Review, status and updates
Review is proportionate to consequence and complexity. It may cover sources, calculations, quotations, regulatory context, terminology, conflicts and the link between evidence and conclusion. Articles may identify evidence or method status to help readers understand the level of support. Material changes are handled under the Corrections process.
7. Limitations
No method removes uncertainty, bias or changes after publication. Readers should consider their organisation, jurisdiction, controls and data before applying a conclusion. Finance Circuit content is informational and is not a substitute for professional financial, investment, accounting, tax or legal advice.