According to Kerry Vice President Marketing & Strategy APMEA Parth Patel, Trendspotter was developed in order to both launch F&B products ahead of trend as well as to reduce the lead time for innovation by up to seven months.
“Imagine, if as a branded food and beverage player you are able to launch a new product ahead of its trend. Basically, you are creating a runway for yourself and most probably you are going to be the only player, hence chances of success are high,” he told FoodNavigator-Asia.
“We aim to reduce the lead time for innovation and the chances of failure of new product launches, while increasing the probability of success by focusing on those trends which are emerging and not yet mainstream.”
The need here comes from high rates of recorded failures – a recent Nielsen study found that 85% of new product launches fail within 18 to 24 months, driving high burdens in terms of cost, time, and effort for all major F&B companies.
“[This] AI-enabled tool [aims to tackle this] as it can predict, with a high degree of probability, soon-to-be popular flavors and ingredients across 60 countries,” Patel said.
“Trendspotter took close to two years to develop together with IBM Watson, and is based on the premise that most innovation happens outside-in - it starts from the fringes (local bars, pubs, cafes, restaurants, street side stalls), moves inwards to food service and finally into the retail packaged sector.”
In accordance with this, the tool works by capturing data from various sources such as Instagram, Google, Quick Service Restaurant (QSR) menus and other social forums, then making predictions based on it’s the proprietarily developed AI algorithm to determine the trends.
A Chinese version of Trendspotter is on its way with a focus on Chinese firms so as to better capture the APAC market as a whole, as is an updated version dubbed Trendspotter 3.0.
APAC trends
Although Patel declined to reveal full details of product brands launched in APAC using this tool due to existing confidentiality agreements, he said that savoury snacks had been one of the major areas of focus so far.
“As a case study example, we recently collaborated with a very large snacks player in South East Asia using this tool, [who] wanted to launch a flavour extension of an existing popular snacks (crisps or chips) brand,” he said.
“We leveraged insights from Trendspotter to identify emerging flavours and suggested those for further exploration. A typical process such as this one might take four to six weeks, but with the help of Trendspotter we were able to revert back with product concepts within just five days.”
“The entire process from ideation to commercialisation took less than two months, [up to seven months faster than normal, as the] typical time for such projects could take six to nine months. [One of the main benefits of moving at lightning speed here] was that it helped them to gain market share with the new launch.”
Other major trends that have been identified in the APAC region using Trendspotter revolve around a ‘multisensorial’ aspect, going beyond taste to cover all senses including sight and touch.
“Major trends we have identified in the APAC region include beverages that indulge all five senses, for example avocado coffee, blue tea and nitro cold brews,” said Patel.
Blue tea undoubtedly appeals to the visual sense for its bright colouration, whereas avocado coffee – one of the most popular variants being a vegan avocado latte dubbed ‘avolatte’ which is served in an avocado shell – appears to call to both sight and touch, and is gaining exceptional popularity in Australia.
“Asia is increasingly seen as a trendsetter and there are many innovations that are now going west from here. Think of fermented drinks like kombucha, cheese tea, bubble tea, the entire Korean food revolution,” he added.
Trendspotting challenges
Patel added that this tool is targeted to intelligently spot trends across all F&B categories, both in packaged goods and food service, and highlighted that in doing this, one of the major challenges the team faced was to train the AI and develop the right algorithm.
“[It was key to develop the algorithm and train the AI so it produced consistent and highly reliable predictions, [and] patience is key in working with any AI,” he said.
“You have to teach it to be intelligent and aligned with your expected outcomes; AI learns really fast so when you teach it the right things you move along quickly, but it moves equally fast when you teach it the wrong things.”
Other areas requiring much focus included data management and UI design – all of which carry monumental importance as Kerry’s Trendspotter is not the only AI-based tool in town with this focus.
Singapore-based start-up AI Palette also does essentially the same thing, predicting F&B trends using AI, but has the supposed advantage of being ‘language agnostic’, and so works with all sorts of non-English languages.
Though currently running on a smaller scale, AI Palette recently secured US$1.05mn in funding and has plans to scale up, so time will tell whether this will give Trendspotter a run for its money.