Personalised flavour in food industry overview
- Food industry increasingly uses data to target flavours by demographics
- AI enables precise segmentation moving beyond traditional one size fits all
- Taste preferences linked to age, gender, culture and geographic location
- Companies balance personalised products with operational efficiency and scale constraints
- Future lies in advanced segment targeting not full individual personalisation
We live in an increasingly personalised world. From algorithms recommending films and music to AI assistants tailored for a single individual, it is now common to see products and services moulding themselves to personal preferences.
In food, personalisation is in full swing. Personalised nutrition is popular with consumers, and industry is increasingly interested in personalised taste.
Food companies want to know how different consumers will react to different tastes, so they can tailor products to smaller and smaller groups of people.
Shift away from ‘one size fits all’
How companies look at flavour is changing. Market segmentation is becoming more precise, and industry is zeroing in on clearly defined consumer groups using vast datasets.
The shift away from a one-size-fits-all approach to food is accelerating, says Anne Berends, R&D director for UX/PX at consumer product experience and sensory services for EMEA at ingredients company Cargill.
This is being driven by “more individualised definitions of what health and what enjoyment today really mean.”
While brands have always segmented, “what has changed most is precision rather than intent”, says Bernard Lahousse, co-founder of AI company Foodpairing, which helps FMCGs predict trends and has partnered with PepsiCo and Nestlé in the past.
Technology advances can make predictions more precise and more scientific, shifting personalisation “from a marketing story to a measurable engineering problem“.
How different demographics react to flavour
Consumer preferences on taste can be linked with increasing accuracy to different demographics.
For example, age is a key factor in determining food preferences, says Foodpairing’s Lahousse. Younger consumers, aged roughly between 21 and 39, have far more narrow tastes than older people, whose preferences are more diverse.
With older consumers, “you need a wider portfolio to achieve equivalent reach” because “you are serving a more diverse set of tastes.”
“Personalisation has shifted from a marketing story to a measurable engineering problem.”
Bernard Lahousse, co-founder of Foodpairing
While certain flavours sometimes correlate with demographics, it’s not so much about a particular demographic’s inherent preferences, and more that they are familiar with certain flavour profiles, suggests Cargill’s Berends.
For example, Cargill has found that men have a preference for dark, roasted and bitter profiles in chocolate. But this preference is only slight – in fact, 2% have a higher preference than women.
On its own, says Berends, this difference is too small to be significant. However, it led to Cargill exploring how it could link to other dietary habits, such as men also drinking more bitter and smokier drinks.
“In the end, it’s not so much the gender that we see as driving eating preferences, but rather the exposure to specific flavours and foods.”
Other demographic categories are also linked to taste preferences.
Targeting tastes by geography
Geography can shape tastes. Some taste preferences are innate, says Berends – a newborn baby, for example, will react well to sweetness and reject bitterness – but other flavour preferences are trained by cultural context.
For example, UK and French consumers usually reject chocolate containing butyric acid, which is commonly used in American chocolate. This is because these consumers are not familiar with it.
“We don’t want to guess [what consumers want]. We want to be sure from the start.”
Anna Berends, R&D director for UX/PX at consumer product experience and sensory services for EMEA at Cargill.
While the majority of consumers are interested in new and novel flavours, valuing the “cultural heritage” of consumer demographics remains important, because familiarity is an essential need.
Global and local food preferences must be weighted against each other, says Foodpairing’s Lahousse. In some markets, such as Sweden and the US, local preferences align strongly with global ones. In others, such as the UK, there is a stronger divergence.
This framework also works at a regional level. Preferences that dominate at a national level in the US, for instance, don’t even show up for certain regions.
To understand consumer preferences, companies must come to grips with how demographic and local preferences intersect. Localisation and demography data cannot be looked at separately, Lahousse stresses. “You need a joint model that holds both simultaneously.”
The role of AI in personalisation
A big part of personalising food is crunching the numbers: making sense of the consumer data to discover what preferences are associated with which demographics and geographies.
AI can streamline this process, because as in many other fields, it can work much faster than humans, explains Berends.
It can also collect disparate data that, in the past, would have been in silos, and put it into context.
AI can bring together data on the taste of foods, consumer feedback data, and performance data. It can match these up: for example, it can match a taste profile to consumer feedback.
AI can even allow companies to tailor their formulations to different situations or moods: the formulation of an ice-cream a consumer buys whilst shopping during the day may be different from one that they eat in front of a film in the evening, for example.
The aim is for the company to be able to anticipate consumer preferences, rather than simply to react to them.
“We don’t want to guess,” says Berends. “We want to be sure from the start”.
The precision that AI can achieve is immense. Molecules can be “mapped to perceptually calibrated sensory descriptors”, says Lahousse, which is what “distinguishes genuine prediction from correlation noise”.
Is personalisation the future of food?
So, with all the focus on meeting the demands of particular demographics, could this lead to a permanent shift in how food is developed?
The industry is moving towards “more precise segmentation”, says Berends, rather than full individual personalisation.
Personalisation must be balanced with operational efficiency; if companies were to make a different formulation for every consumer, there would be no economies of scale. Nevertheless, advanced segmentation enables industry to target consumer categories more precisely.
Foodpairing’s Lahousse agrees. “The realistic near-term endpoint is not one-to-one personalisation. It is intelligent segment-level precision — moving from 4–6 product variants to 15–20, each backed by a validated consumer model rather than category intuition.”
So personalisation can only go so far. But while targeting flavour preferences for individual consumers is not always viable, targeting demographics is getting easier and more precise.
