Product Development

You Need to Change Your Sensory Analysis. Here’s Why

At Gastrograph, we agree that sensory analysis is a useful tool for new product development. In fact, we'd argue that understanding how people perceive your product is essential to making smart product decisions.

Sensory analysis, together with consumer research, is currently considered by the industry and researchers to be one of the most useful tools at the different stages of new product development....”

At Gastrograph, we agree that sensory analysis is a useful tool for new product development. In fact, we'd argue that understanding how people perceive your product is essential to making smart product decisions.

Sensory analysis involves defining the properties (texture, flavor, taste, appearance, and smell) of a product. It’s the foundation you build your product development on. You don’t want a shaky foundation.

But conventional sensory analysis is fundamentally flawed. Gastrograph AI offers an alternative that’ll help you meet the challenges of 21st-century product development. 

What Does Conventional Sensory Analysis Look Like?

Almost 70% of food companies — including Vimto, Goodfellas, and Subway — use sensory analysis as part of their new product development. There are a variety of ways to conduct sensory analysis, and scientists are developing new methods all the time. 

For example, there’s descriptive profiling, free-choice profiling, and newer methods like check-all-that-apply (CATA). Each method has its pros and cons, so companies choose the one that best suits their goals and budget. 

Descriptive profiling is the classic method. Highly trained panelists identify the attributes that are in a product and generate a consensus profile — the attributes that the majority of panelists agree on end up in the profile. Companies consider this as the most reliable way to get a detailed picture of your product, but it’s expensive and time-consuming. 

What’s Wrong with Conventional Sensory Analysis?

By sticking to conventional sensory analysis methods, you’re jeopardizing the success of your products.


Calibration Trained Experts Skew Your Results

In most sensory analysis methods, reference trained experts identify the flavor attributes in a product. You treat people either as sensors (who identify the attributes in a product) or consumers (who give you their opinion about a product but don’t get to identify flavors). This process doesn’t give you accurate or useful tasting data. 

Reference trained experts calibrate their taste buds by learning the “standard” of different flavors. So if we decide Smucker’s blackberry jam is our standard for blackberry, the professional tasters evaluate the blackberry-ness of all products by comparing them to that specific blackberry jam flavor. 

The training process means the expert tasters have a greater ability to identify flavors than the average consumer. That also means they don’t represent the average consumer.

It takes a lot of time, money, and effort to do this training. When you use expert tasters, you rely on their ability to act un-human: to resist being opinionated and to avoid making mistakes. But tasters are humans; they’re not sensors, and “treating us like sensors for the most part does not make us better sensors… humans are very opinionated, and they're going to find ways to insert their opinions into your results,” explains Jason Cohen, founder and CEO of Gastograph. 

By using expert tasters for your sensory descriptive analysis and then getting consumers to rate their preference, you separate perception from preference. Your data sets are disjointed.

You also limit the terms that consumers can use to describe their experience of your product, so you don’t get the full picture of what they taste. When this happens, consumers might offload their opinion of flavors that the experts don’t include on the attributes list onto other flavors, so the results of your preference analysis are inaccurate.

 Static Data Wastes Your Money 

Data from conventional sensory analysis is only valid in an isolated way. Companies have to choose between different methods, so there’s variation in the testing results. You can’t use the data to make comparisons with the products’ results from other tests. 

You also can’t reuse your data or mine it for new insights. Each set of analyses is valid only once, which makes the cost of the tests you do very high. 

When your testing isn’t standardized, you don’t have a standard profile of all the products that you can compare. It would be like trying to compare our exam scores when I took a multiple-choice test and you wrote an essay. 

Even if you stick to one method, you still have a lot of issues. Tasters generate a list of attributes to describe the product rather than considering the same attributes every time, so you can’t compare profiles across products, and you might miss tasting data. 

Consider marine flavors in beer. Most beers don’t have marine flavors, so they wouldn’t appear on an attributes list. The problem is, mis-brewed beer does contain marine flavors — it tastes like iodine. So when consumers are rating their preferences for the attributes on the list, they have no way of communicating that the beer is flawed. 

By using traditional descriptive profiling, you don’t give consumer tasters a way of communicating a full profile of flavors — ”where [they’re] supposed to be there, or even not supposed to be there,” says Jason. 

How Does Sensory Analysis Work in Gastrograph AI?

Tasting is a wonderfully complex experience. We’ve developed a system that navigates this complexity while giving you actionable data to inform your product experiences. 

Very Human Tasters Give Accurate Product Analysis

Rather than training humans to calibrate their taste buds as in conventional sensory analysis, we’ve calibrated the Gastrograph system to understand how humans taste. 

At Gastrograph, we use a type of free-choice analysis for our tastings. People are free to describe flavor in the way that makes the most sense for them in any language they want. Then our system uses natural language processing to understand what people taste based on what they say.

It’s like how Google Search makes sense of the words you use in context to deliver search results. Because the labels tasters enter fall under the 24 attributes, the tasting results are statistically useful, and we can compare flavor profiles. 

In conventional product testing, there are separate tasting rounds — sensory analysis with expert tasters and then preference analysis with consumers. In the Gastrograph system, one taster does everything. 

Consumers taste a product, fill out a full flavor profile by describing what they taste, mark the intensity of each flavor, and then rate their preference for it. People are free to record what they really taste when they try a product, so we avoid the risks of flavor offloading, disjointed data sets, and incomplete flavor profiles.

Evergreen Data Informs Smart Product Decisions

In the Gastrograph system, we have a standard way of creating flavor profiles. We're able to directly compare the flavor profiles of different products and repeatedly mine those profiles for new insights. 

When a taster brings a new product onto the Gastrograph system, they select from 24 attributes and then add their own labels to describe what they taste. The 24 attributes on the system cover the “gustatory flavor space,” so there’s nothing a consumer could taste during a standard tasting that would fall outside of those 24 categories, explains Jason.

Tasters add labels under the same attributes every time they try a new product — instead of just considering the attributes they’re “supposed” to taste, as they do in a conventional tasting. Because tasters aren’t primed to notice any particular flavors, you’re more likely to get a full and accurate picture of what they perceive. 

The objective flavor profile for all products takes the same format. We compare different products and run variations of each product through our system to predict consumers’ preferences. 

The green attributes had a positive impact on preference; the red ones had a negative impact

Brands often use conventional sensory Product Quality Benchmarking to compare their products with their competitors. It’s expensive and time-consuming because, for each comparison, you have to run a new tasting. 

But Gastrograph’s tasting data is reusable. After you’ve brought a product onto the system once, without re-sampling, you can use it as a benchmark and compare multiple different products against it.  

We also “translate” perception data onto different demographics. Let’s say we brought a product onto the system in Paris. Without running any more tastings, we can translate consumers' perceptions of that product to predict how much consumers in New York, Beijing, or London will like it. We’re able to do this for more than 30 different countries, and we’re constantly growing our sensory database to include more regions. 

Future-Proof Your Product Testing

We’re continuously updating our database with new tasting information. Gastrograph AI models how consumer preferences change over time, so we’re able to make predictions about how much people will like your product in the future. 

In comparison, the inflexible data you get from conventional analysis is only valid for a short amount of time. It has no way of accounting for the fact that consumer preferences change, so it has a short use-by date. 

If you're sticking to traditional sensory analysis, you're choosing to get left behind.

Ready to explore how you can use Gastrograph to future-proof your products?



Similar posts

Stay up to date with Gastrograph AI

Be the first to know about company updates and industry-related news, straight from our internal subject matter experts.