Before interpreting a change, check whether the two figures describe the same business, time span, currency basis, and definition.
A percentage change is only as useful as the comparison underneath it. Two correctly extracted figures can still produce an unhelpful conclusion if they describe different periods or definitions. The comparison needs its own review before it becomes part of the argument.
This is a small discipline with a large practical benefit: carry the label, unit, reporting period, and source alongside the value. When those fields travel together, a reviewer can see whether the proposed comparison is appropriate without reconstructing the original table.
Start with the time span
Separate a quarter from a fiscal year and a point-in-time balance from an activity measured over a period. Do not place them in the same comparison simply because the numbers appear in the same release. Check whether both observations cover equivalent durations.
Keep the fiscal calendar visible. The month a report is published does not determine the period it measures. If the research question is about sequential change, identify the earlier period that actually belongs in that sequence.
Read the definition beside the number
Similar labels can describe different measures. A product, a reporting segment, and a broader commercial category may overlap without being interchangeable. Preserve the definition the company supplies and note when that definition changes.
The SEC’s guide to company reports points readers to the financial statements and accompanying notes. Those notes are part of the research material, rather than optional context to remove after extracting a figure.
Check units and presentation basis
A figure shown in millions should not be compared directly with one shown in billions without a documented conversion. Do the same for currency and any alternative presentation basis. Keep reported and adjusted measures distinguishable, and follow the relevant explanation or reconciliation before interpreting the difference.
A clean table in a brief should make these choices visible. The reader should not have to guess whether a growth rate is reported, calculated by your team, or presented on another basis by the company.
Treat recasts as new comparison evidence
When a company changes its segment presentation or revises a comparative table, use the information appropriate to the question. Do not casually combine a figure from an earlier report with a later figure whose comparison basis has changed.
Record which version of the comparative data the brief uses and why. If the team previously published a note on a different basis, retain a short explanation of the change so the research history remains understandable.
Keep calculations reproducible
For a derived measure, record the inputs and formula. A reviewer should be able to recreate the result from the cited values. Retain the underlying figures at their available precision and choose display rounding only at the presentation step.
Before distributing a comparison, ask another researcher to read only its labels and source references. If they cannot identify the two periods, the measure, and the calculation basis, the comparison is not ready for the brief.
Gatani’s research automation approach makes these definition and review steps part of the output design. The objective is straightforward: let the researcher debate the business interpretation rather than discover a hidden mismatch in the inputs.
This editorial collection was published on . Archive dates organize the series.