Questions That Matter to Understand Your Audience in the AI Age

The world of social, opinion and market research is changing rapidly. For decades, the foundations of demographic segmentation—age, gender, income, and region—have served as the backbone of audience segmentation and analysis. Yet as societies evolve and behaviours converge, these traditional indicators have become less predictive of attitudes, lifestyles, and decision-making. The Esomar Demographic Committee, led by a collaboration of researchers from across the research industry, has launched a global research initiative to reimagine how we define and measure the core building blocks of demographic identity in the AI age.

This initiative, now in its second wave, seeks to establish globally consistent, reliable, and future-proof measures for the most common demographic questions asked in surveys around the world including Age, Gender, Working status, Education, Income/Wealth, Occupation, and Sustainability and has a wider ambition to develop a unified framework for assessing Social Class—a measure that reflects not only income or occupation but also social standing, opportunity, and attitudes toward life.

Esomar’s Global Demographic Research Programme

Wave 2 of the global demographic research programme fielded responses from 20,000 individuals across 20 countries. Each wave explores methods to refine how researchers measure key demographic dimensions in surveys. Having covered Age, Gender and Working Status in Wave 1, this phase focuses on the remaining four pillars: Education, Wealth, Occupation, and Sustainability—areas that underpin most existing social classification systems.

The initiative builds on the premise that demographic measures should be consistent across cultures while remaining sensitive to national context. It also aims to integrate behavioural and attitudinal data, ensuring that demographic variables remain meaningful in a world increasingly defined by digital and cultural fluidity.

Measuring Education on a Global Scale

Education remains one of the most universally used demographic questions in research, yet it is also one of the most inconsistent. Different national systems, changing qualification structures, and generational shifts make cross-country comparisons fraught with error. Moreover, many respondents overclaim qualifications such as degrees, making the data unreliable for international benchmarking.

The Esomar committee’s solution is a modular approach that reconstructs educational attainment through a sequence of simple, factual questions. Rather than asking directly for the ‘highest qualification obtained,’ the method pieces together educational history from key milestones: the age respondents left school, whether they pursued further education after school or not, what type of further education they undertook, and whether they obtained qualifications. This allows researchers to infer level of attainment more accurately and consistently.

Testing has shown that the new approach significantly reduces overclaiming of degrees. Despite involving several questions, it typically takes under 30 seconds to answer the revised cluster of questions.

The total number of years spent in education also performs well as a short-form proxy for educational level, offering a valuable fallback for quick surveys. Guidelines for international application are planned for publication by Spring 2026.

Measuring Wealth Beyond Income

Wealth is perhaps the most complex demographic to measure. Income alone tells only part of the story—it does not account for savings, assets, spending power, or financial stress. In 2024, the Esomar committee explored multiple dimensions of wealth: formative (income, assets), reflexive (possessions, expenditures), subjective (self-assessed status), and predictive (education, age, working status).

A key output was the development of a ‘meta-wealth’ concept, combining income, equity, and spending ability while accounting for financial strain. Analysis by students at the University of Southampton revealed that the link between income and lifestyle behaviour often depends on both the willingness and the ability to spend money. This insight underpins the design of a new hybrid wealth measure.

The goal set for Wave 2 was to refine the approach further, adding new elements such as ownership of investments, indicators of financial strain, and expenditure categories including pensions.

The early findings suggest that income remains an important component but works best when paired with contextual indicators of spending freedom and security. A comprehensive set of Esomar guidelines on wealth measurement is planned for spring 2026.

Rethinking Occupation Measurement

Occupation classification has become one of the biggest challenges in global demographic research. Each country maintains its own classification system—many of them outdated or inconsistent with modern labour markets. Respondents often struggle to find their exact role on long lists, leading to poor data quality. The UK’s official list, for example, contains over 1,000 jobs and has not been fully updated this century.

To tackle this, the committee created a universal job classification framework containing 230 roles grouped across 20 industries. The list was developed iteratively across five waves of research, cross-validated against open-ended job data, and reviewed by three leading AI models—GPT-5, Claude 4, and Gemini 2.5—to identify omissions and inconsistencies.

Because the full list is too long to display in surveys, a set of four pre-filtering questions was introduced to narrow down options based on workplace type, interaction level, and need for qualifications. Additional post-clarification questions capture job responsibility, company size, management scope, and decision-making authority. This approach achieved an 88% job-finding efficiency rate, a major improvement on traditional methods.

Perhaps the most radical finding is that once pre- and post-filter questions are combined, they explain much of what the job title itself would tell us. This raises a provocative question: if we can already infer someone’s industry, responsibility, and work context, do we really need to know their precise job title at all?

Audience Segmentation for the AI Age

The committee’s work has also expanded beyond traditional demographics to question their ongoing value. In a recent experiment testing 300 binary segmentation questions, classic variables such as age, gender, education, and marital status ranked among the weakest differentiators of opinion. Age, for example, ranked 150th as a predictor of attitude differences; gender ranked 180th; and education 220th.

By contrast, lifestyle, attitudinal, and behavioural questions—such as views on owning luxury products, whether someone uses AI tools, belongs to a gym, or follows celebrity culture—proved far more discriminating. Even simple questions about technology adoption personal values, or car ownership, predicted broader behavioural patterns with more differentiation than most of the common demographic questions asked in surveys.

This suggests a paradigm shift: to understand people, we need to move from static demographics to dynamic psychographics and life-context indicators. Potential new segmentation dimensions include worldviews (community vs. individualism), trust and optimism, social values, environmental attitudes, digital literacy, and consumption styles. The art of segmenting in the AI age is to find variables that are predictive, novel, reliable, and globally comparable.

A Call to Action

The Esomar Demographic Committee’s work is setting the stage for a new era in social measurement. The old demographic categories will not disappear, but they must evolve. By combining factual indicators such as education, wealth, and occupation with richer attitudinal and behavioural signals, researchers can build a more complete, truthful picture of society.

The next step is to formalise these new methods into globally endorsed standards—tools that all researchers can use with confidence, whether they are measuring a global brand tracker, a social attitudes study, or an AI-driven digital twin simulation. Esomar invites collaboration from across the research community to refine and test these ideas further.

In the AI age, asking the right questions matters more than ever. Our challenge is not just to keep up with social change, but to measure it meaningfully.