Using research figures without overstating the findings

The total sample is not the base for every finding

Edelman’s 2026 Trust Barometer surveyed 33,938 people online across 28 countries between 25 October and 16 November 2025. The methodology places the study in context, but its global sample does not establish the reliability of a French result or a particular subgroup.

Check the population behind the percentage you intend to use. A finding from one country or part of the sample should not be presented as the view of everyone surveyed. The number on the report cover cannot settle that question.

A US survey describes US respondents

The CMO Survey 2026 reports 308 responses from 2,111 invitations, a 14.6% response rate. Fieldwork ran from 7 to 29 January among marketing leaders at US for-profit companies. The report also gives response counts by question.

The findings can inform a discussion in France, but they do not describe French companies. The response rate warrants looking at who answered; it does not by itself quantify bias or invalidate the study.

Look beyond the report landing page

Salesforce’s announcement for its tenth State of Marketing report specifies 4,450 responses from marketing decision-makers, collected between 8 October and 17 November 2025 across several world regions. The report’s promotional page provides less detail.

Following the original announcement helps recover that context. A regional claim still requires the relevant response base. Without it, a statement about European marketers may be broader than the available evidence allows.

Keep a forecast labelled as a forecast

In February 2024, Gartner forecast a 25% decline in traditional search-engine volume by 2026. The release records the forecast; it does not observe the decline or publish the detailed calculation behind it.

A citation should therefore retain the author, date and the word ‘forecast’. To say the decline happened would require later data measuring the same volume. Advertising revenue would not answer that question.

Use the source to support a judgement

A study is useful when it clarifies an uncertainty for the business. The question therefore precedes the figure: is the aim to understand an expectation, assess a practice or compare a result? A source can strengthen a conviction, limit it or prompt reconsideration. Looking only for a percentage that makes a presentation persuasive loses that function.

Check your conclusion too. A stated preference for expert content does not establish that it caused a purchase. If the source supports a narrower claim, narrow the wording. When publishing company research, give the teams adapting the findings the same definitions and limitations.

Communications leaders should be able to defend the step from observation to decision. Population, period and measure remain beside the figure, with detail accessible in the linked source. Readers can then examine the company’s reasoning and what it means for its market, without accepting an institute’s reputation in place of an argument.

Before reusing a statistic, complete one sentence: what does this source measure, among whom, where and when? If the headline expands any of those terms, correct it or provide further evidence. This tests fidelity to the source independently of how interesting the number seems.