Data and machine learning are increasingly being used by food and beverage brands to develop new products and to identify the flavours that will work best in new markets.

An ESOMAR paper observes that a common practice is to launch one basic variant, together with two or three flavours, in a given market, with market research then executed to identify the flavours with the most potential.

“But this comprehensive research costs time and money,” note the authors, Sjoerd Koornstra of The House of Insights (Netherlands) and Wim Hamaekers of Haystack International (Belgium).

“Moreover, some markets are too small/underdeveloped to allocate research budget to, and are therefore neglected research-wise,” they add.

But by tapping into existing data sources, they were able to uncover a hidden layer of insights and predict the likelihood of a flavour succeeding in any given beverage market in the world, they report in their paper Demystifying machine learning: accelerating innovation by exploiting what is already known.)

The pair developed a database of 49 features, covering everything from country, existing flavour segments and previous flavour introductions to category volume and growth, with the aim of giving a potential score for each flavour in each country.

“In total, the application was able to give indicative insights on almost 50,000 products,” they report. “To collect similar results with traditional research methods would cost easily over €100m and years of fieldwork.”

In China, Alibaba is using its vast trove of consumer data collected from people shopping, searching and sharing on its various platforms to perform a similar role via its Tmall Innovation Center.

“We can see where there are blank spaces and unmet needs in the market,” Duan Ling, Tmall’s director of brand marketing and head of the innovation center, told Bloomberg recently.

That ability led directly to Mars creating a Spicy Snickers bar containing Sichuan peppercorn, based on the insight that the same people who buy chocolate also like spicy snacks.

Not only was the timeline for new product development greatly shortened, but Ian Burton, China president of Mars Wrigley Confectionery, reported that the launch helped meet a goal of having new products account for more than 10% of revenue.

“The age-old tradition is that 90% of innovation fails,” Burton said. “This helps us bring the rate of failure down.”

Sourced from ESOMAR, Bloomberg; additional content by WARC staff