How Ingredient Informatics Are Revolutionizing the Way We Develop Foods

Something we have been talking about at Lux lately is converting weak signals into strong insights. In other words, how do we as researchers turn the aggregation of anecdotal points of evidence from interviews and market observations into strong predictors and recommendations about a space of interest? Recently, one space those weak signals have been culminating in is what we call "ingredient informatics" (analogous to materials informatics). The concept concerns approaching food ingredients like materials; using data, analysis, and artificial intelligence (AI) to reduce costs, improve outcomes, and accelerate the development of new food and beverage products. Here, we'll outline the weak signals that generated the concept of “ingredient informatics” and how they are now materializing into strong insights about when and how CPGs should look to transform their approach to new food and beverage product development.    

Weak Signal #1: The rise of the broader big data analytics and AI space has been on our, and our clients', radar for some time, but in July 2017, we recognized that a whopping $175 billion had been injected into the area, with nearly 75% of those investment deals happening in the five years prior. With all the coinciding hype, many of our clients, especially those involved in manufacturing in one form or another, were (and still are) being approached by dozens of analytics providers, with each pitching how their novel AI algorithms can enable new applications, save millions on operational costs, and the like. In an effort to help navigate this haze, we looked at some realistic use cases for AI, what applications manufacturing industries were pursuing, and what capabilities clients should look for in an analytics platform. At that time, we actually observed no adoption of AI-fueled analytics for new technology and product development within the food and beverage industry. However, major adoption was occurring in adjacent (albeit distantly adjacent) markets: pharma and chemicals and materials (i.e., AI for drug discovery and materials informatics, respectively)         

Weak Signal #2: Around the same time, not only were we seeing the adoption of materials informatics, but we were actually seeing it deliver value through several case studies where players were leveraging the technology. 

Here are two examples: 

·       Citrine Informatics helping a Fortune 1000 engineering design company save more than $25 million in less than three months to determine the mechanical properties of 6,000 materials without physical experimentation

·       Nutonian saving a Fortune 100 industrial company two months of time and upward of $500,000 to streamline the number of forged parts required to validate turbine disk performance.

It was case studies like these that made us select materials informatics as one of the four key digital technologies for chemicals and materials players to exploit in order to meet increasing efficiency demands.

Weak Signal #3: In parallel with the signals listed above, we had begun to recognize a troubling pattern in the food industry of either dismissing digital tools as "not for me" or diving too deep into using digital tools without legitimate justification for doing so. Many CPGs were continuing to use sensory panels, expensive and deeply experienced formulation scientists, and closely guarded "secret" recipes to develop their products, often with unsuccessful results–along the lines of the now-infamous noisy Sun Chips bag. Others were using overengineered solutions like's blockchain-for-ag approach that has since been panned by most would-be users. There were precious few developers approaching digitalization in the food industry with what we saw as the "right" approach. We came to recognize this as a signal of a major future problem for the food and agriculture industries, which started us down the path to help teach our clients how to engage more productively with the digital world. From this recognition, the concept of "ingredient informatics" was born.

Lux Research Ingredient Informatics

During last year's Lux Executive Summit, we introduced the concept in a presentation titled "You need a new strategy for food innovation," which laid out a roadmap for CPGs and others looking to build their own ingredient informatics approaches to food innovation. At the time, the only real examples of such efforts we could uncover were NotCo and IBM's Chef Watson. Since then, however, we have seen a significant uptick in market activity around companies that are exemplifying the concept of ingredient informatics, bringing new players onto our radar. We highlight the recent events tied to these companies below, segmenting the companies into two buckets: "brand owners" (those using AI-based data analytics to create their own or co-developed products) and "service providers" (those offering AI-based data analytics as a software service to CPGs).

These events highlight how the concept of "ingredient informatics," that was brought about by observation of the previously mentioned weak signals, has actually come to life. Given the strong need to reinvent new food and beverage product development, coupled with these recent events, we expect the momentum around ingredient informatics to continue and eventually replace traditional approaches. However, the ever-crucial question still lingers and comes in two parts: What should I do about this and how? As ingredient informatics is presenting itself as a way to rapidly develop novel food and beverage products, we'll make the same recommendation we did concerning materials informatics back in 2017: Invest now. As stated then, digitally driven technologies move faster than physical product innovation; now is the time to make these investments to secure privileged access to the tools of tomorrow. To the latter part of the question, which is the trickier part, we recommend utilizing the same framework we provided back in 2017 for selecting an analytics provider. At least looking at the "service providers," there is minimal differentiation among the lot. Thus, using our framework, first, identify the most desired platform features. Then send RFPs to potential vendors with the list of desired features, and document and rate the vendors. Finally, implement POCs and track long-term ROIs.


Questions? : 
Email questions to