Sensory White Spaces: A Path to Better Innovation

Written by Bernard Lahousse - 02.03.2023

What is the unique selling point of your new product? In food and beverage innovation, the answer may lie in identifying sensory white spaces that allow your brand to deliver a stand-out consumer experience. Digital tools can help you achieve just that. 

What is white space?

Some believe a white space is a unique market segment where there is no competition. Others consider white spaces to represent entirely new markets. But when we talk about white spaces at Foodpairing, we mean gaps in an already existing market – the opportunity spaces between the products your customer has already purchased from you or your competitor, versus all products that could exist in your market. 

To put it another way, White Space Analysis is a process businesses use to uncover new opportunities in the market. This process requires leveraging your existing sales data to identify the gaps and discover potential consumer needs that aren’t being currently met. 

The White Space Analysis tool is both popular and respected as it can give your business the competitive advantage in a crowded market. Based on existing customer data, White Space Analysis serves as an effective practice to provide you with powerful insights into how to grow your business.

What does Foodpairing do differently?

If the analysis of white space is already a popular part of the innovative process in your portfolio management, what new approach does Foodpairing bring to the table? Traditionally, white space mapping focuses on sales data. We add an additional layer, through our AI tools we are able to predict a liking score for all possible product variations within the flavor territories selected.

This is important. There might be a very good reason that there aren’t any products in a white space: there may be no demand. For many variations, you just don’t know how much your new product will resonate with consumers. 


white spaces innovation graph


That is why Foodpairing has developed the tools to predict how much your target consumers will like existing and future product variations. By averaging the product likings per flavor territory combination, you get an indication of how popular that flavor territory will prove.

By combining white spaces based on sales data and predicted liking data, you can detect the highest liked flavor territories where the offering does not match the demand in the market.


How to predict whether products not on the market will be liked? 

To predict how successful a new product will be when it launches, Foodpairing developed two models.

  • The first is a model for trial, forecasting purchase intent.
  • The second is a model for liking, used to determine the potential of any white spaces that are detected. 

Foodpairing’s liking model selects digital twins of your target audience and, for each one, predicts how much that consumer will like virtual recipes within the flavor territory. Digital twins are built based on real consumers that fit your target audience. 

Foodpairing has been digitizing consumers and food for more than 10 years. By combining data from the Foodpairing tools like FlavorID, chef apps and public data like recipes and social media, Foodpairing has successfully digitized consumers with a diverse demography.

flavor beetroot trend graph

flavor beetroot trend graph


Foodpairing’s technology is built on the chemical flavor understanding of ingredients and the consumer preferences to the individual aroma compound. As such, Foodpairing does not need to have data on all ingredients to be capable to predict the trial and liking of ingredients.


How is the liking model validated? 

To validate the liking model, we compare its findings against real-world data coming from a consumer test (CLT, HUT). This data mostly comes from our clients. It can be historical data but as long as the products are still available for Foodpairing to analyze them.

Another option is that Foodpairing initiates a CLT or HUT using commercial products on the market. The minimum number of products is around 20. Validating on a lower number of products does not provide a robust correlation metric.

For line extensions a minimum correlation of 70% (MAE of 8%) is needed to start a project. For core product upgrades a correlation of 85% and MAE of maximum 4% is required to make any useful predictions. 


Real world use case: Philadelphia-Milka cream cheese 

Foodpairing’s approach to white spaces has already delivered successful NPD to retail shelves. 

When approached by Kraft Heinz and Mondelez International, we were able to suggest an innovative way to stretch the Philadelphia and Milka brands

Looking at what ingredients match the unique aroma profile of Milka chocolate and combining this with product mapping data and learnings from trial and liking models, we identified untapped market opportunities, application areas and eating moments. 

Focusing on a growing category within the breakfast area – low-fat chocolate-cheese spread – the companies were able to launch a new product that combined Milka’s sweet taste profile with the creaminess of Philadelphia cream cheese. 

The results were overwhelmingly positive, 74% of consumers said the combination had a ‘delicious taste’ and the new SKU generated sales of £9.26m in its first year on the UK market.



The Power of AI-driven NPD

Instantly evaluate millions of auto-generated new flavor concepts, product variations and formulations against a digital twin of your target consumer anywhere in the world with high predictive accuracy. The Foodpairing platform empowers diverse teams to make smarter decisions, faster, thus creating value across the entire business. Learn more about our technology here.

What flavor white spaces should you invest in first?