ESOMAR 2024: Ipsos puts AI empathy and scent data on trial

Ipsos has laid out a dense programme for ESOMAR 2024 in Greece, and it works as a preview of the next insight toolkit. Across 9 to 11 September, the firm is presenting sessions on fragrance emotion, synthetic out-of-home audiences, demographic data quality, AI in qualitative research, knowledge graphs and L’Oréal’s use of generative AI.

The conference theme—mind, myth, machine—separates the programme into a useful checklist. The mind sessions dig into psychological and demographic signals. The myth work challenges whether tools such as AI actually deliver human empathy. The machine sessions push synthetic data, knowledge graphs and generative AI into insight workflows.

Scent, self-report and cultural context

The opening session pairs Ipsos with Sol de Janeiro to map emotional response to fragrance across the United States and China. That matters because fragrance buying is rarely a rational like-dislike decision. Emotional associations differ by culture, and a survey question may not capture the layered reasons a scent converts.

For brand managers, the implication is to treat stated preference as useful but incomplete. The method here points toward measurement that goes beyond what shoppers can easily say.

Synthetic journeys and out-of-home reach

Another session simulates more than four million individuals in a virtual France to model how busy streets would be at a national scale. Out-of-home planning has traditionally leaned on panels, historical counts and spot checks. A synthetic population is a model-generated approximation of movement, which can fill gaps where direct observation is expensive or impractical.

The strategic promise is not perfect precision but earlier, broader simulation. It can help a brand test exposure assumptions before committing media budget.

AI’s empathy test

The most emotionally loaded session asks whether AI fosters empathy or leaves qualitative research feeling empty. The self-funded pilots were run in the US, UK and Japan, and the paper is a finalist for the Best Paper Award. That framing matters because qualitative insight has always been built on human listening. If AI moderates, summarises or probes, the risk is that the output feels fluent but not felt.

Brand teams should treat AI as a junior analyst, not as a replacement for lived consumer context. The useful test is whether it helps researchers hear more, not whether it sounds more articulate.

What brand managers should borrow

• Measure emotional response implicitly and across markets, especially for sensory categories.
• Use synthetic or modelled data to simulate reach before buying out-of-home media.
• Pilot AI in qualitative work against a human baseline, then audit for empathy and insight.

ESOMAR’s agenda is not just for research agencies. It signals which measurement methods will soon become standard inside brand teams. The brands that adopt them thoughtfully will be able to separate insight from noise faster than those that wait for the industry to hand them a finished playbook.

Source: Ipsos