Flavour Validation with Sensory Panels: A Worked Example for a Sweet-Heat Snack Seasoning

About this worked example

This is a labelled worked example, not a client case study. The brand, product and brief are illustrative, and no result on this page was measured. It shows how Food Research Lab would plan flavour validation for a new snack seasoning launching in the US and the EU, using the ISO sensory standards for profiling, assessor training and consumer testing.^[1][2][3] Where numbers appear, they are planning calculations from stated assumptions, not data.

FRL is based in Chennai, India. We design sensory studies for US and EU brands and run or arrange the panel and consumer testing through accredited partner laboratories and local fieldwork partners in each market.

The flavour validation brief

An illustrative snack brand wants a honey-chilli seasoning. Sweet-and-spicy, or “swicy”, flavours, with hot honey as a lead example, are named in 2026 trend reports from the Institute of Food Technologists and from Kerry’s 2026 Global Taste Charts.^[4][5] Our flavour trends 2026 analysis covers that direction in detail.

The brief asks flavour validation to confirm three things in both markets:

1. The seasoning is perceived as intended: sweet first, then a clear heat that builds and fades within an agreed time.
2. Consumers like it no less than the leading competitor seasoning, within a pre-agreed margin.
3. It keeps its character to the end of shelf life.

Table 1. Worked-example brief

Label wording, such as whether the chilli or honey note can be called “natural”, is a separate question, covered in our guide to natural flavour labelling under FDA and EU rules.

The challenge: proving a sweet-heat flavour before launch

New flavours tend to fail at the step from concept to proof. A sweet-heat profile is hard to test well for three reasons.

• Heat carries over. Chilli burn builds from one sample to the next, so later samples can seem hotter or flatter than they are unless the test controls rest time, rinsing and serving order.^[6]
• Sweetness masks heat. A change to sugar or honey level can shift perceived heat even when the chilli dose is unchanged, so both need their own attributes on the scorecard.
• Markets differ. US and EU consumers may score the same prototype differently, so liking is measured and analysed separately in each market before any pooling.

Flavour validation therefore needs two kinds of evidence: a description of what the product is, from trained assessors, and a measure of how much people like it, from consumers.

Our approach: designing the validation study

Two tracks run in parallel and meet at one decision (Figure 1). The methods follow the ISO sensory standards; our pillar guide to designing a sensory study explains each test family, and our overview of sensory evaluation methods for food and beverages gives the background.

Trained panel: descriptive profile

Assessors are screened for taste and smell acuity and trained before profiling.^[2] The panel agrees a vocabulary for the product (sweetness, honey aroma, heat onset, peak heat, heat decay, lingering burn, saltiness) and scores each attribute on line scales, three replicates per sample.^[1] Analysis of variance shows which attributes differ between prototypes and benchmarks.^[1] Heat-specific rest breaks and a fixed rinse limit carry-over between samples.^[6]

Consumer test: liking against the benchmark

Each market runs a central location test with recruited target consumers, a 9-point hedonic scale and a balanced serving order.^[3] Every consumer tastes the new seasoning, the competitor benchmark and the brand’s current product, so liking can be compared within each person.

The test is sized as a non-inferiority comparison: is the new seasoning liked no more than 0.5 point less than the competitor? With a one-sided 5% significance level and 80% power, the number of consumers needed is:^[7]

n = (z_0.95 + z_0.80)² × SD² ÷ margin² = (1.645 + 0.842)² × 2.0² ÷ 0.5² = 98.9, so 99 consumers per market.

Here SD is the standard deviation of each consumer’s difference in liking between the two products. Both the SD of 2.0 and the 0.5-point margin are planning assumptions; a live project replaces them with the brand’s own historical data. The calculation also shows why small tests mislead: with 60 consumers and the same assumptions, only a gap of about 0.65 point or more could be ruled out.

Linking the profile to liking

The two tracks are then joined: where liking differs, the profile shows which attributes differ too. With several products spanning the category, preference mapping models how each attribute drives liking; with only two or three prototypes, a direct attribute-by-attribute comparison is more honest than a map.^[6]

Results: acceptance criteria, not measured data

Because this is a worked example, no results are shown. In a live project, the study is judged against criteria agreed before the first session, each at the 5% significance level and in each market separately.

Table 2. Pre-set acceptance criteria (illustrative)

A note on the shelf-life triangle test

In a triangle test, assessors pick the odd sample out of three. With 36 assessors, 18 or more correct answers show a difference at the 5% level, because the chance of 18 or more correct answers by guessing alone is under 3%.^[8] But a non-significant result does not prove the samples are the same. To claim “no difference”, ISO 4120 sizes a similarity test from a small risk of missing a real difference and the largest proportion of assessors who may notice it; both are agreed before testing.^[8]

Development process and iterations

Round one tests two heat levels. If the profile shows heat fading too fast, the formulator adjusts the chilli extract or the sweetness, and the changed version only goes back into testing, with the same assessors and the same design, so any difference comes from the recipe.

1. Round 1: two heat levels. Profile both against the benchmarks and run a small consumer screen.
2. Round 2: one change. Alter heat or sweetness, never both, so the profile shows which change moved liking.
3. Round 3: confirm. Run the full consumer test in each market on the chosen version.
4. Shelf life. Compare aged samples with a cold-stored reference at agreed time points.

Shelf-life sensory checks sit alongside the physical and chemical tests described in our guide to real-time and accelerated shelf-life testing. If an aged sample develops an off-note rather than just losing heat, our article on off-odour and taint recalls shows why that needs a root-cause check before launch.

The launch decision rule and deliverables

Go requires all three criteria in a market. If one market passes and the other does not, the brand can launch where the product passed and reformulate for the other, rather than average the two.

• Trained-panel profile report with attribute ranges and statistics
• Consumer results by market, with the non-inferiority analysis
• Shelf-life sensory results at each time point
• A written go or no-go decision against the pre-set criteria

If the brand wants to use the results in a claim, such as “tastes as good as the leading brand”, the study design has to support that claim; ISO 20784 sets out guidance on the evidence.^[9] Our sensory testing regulations Q&A explains how US and EU claim rules apply.

Lessons for your own flavour validation

• Write the decision rule before testing. Criteria set after the data arrive are easy to bend.
• Give heat its own scales. Onset, peak, decay and lingering burn behave differently and respond to different recipe levers.
• Size the consumer test. A calculated sample size tells you what the test can and cannot rule out.
• Recruit by usage and test each market separately before pooling.
• Do not read “no significant difference” as “the same”. Use a similarity design when sameness is the goal.

Our article on challenges in the sensory evaluation of food covers more sources of bias, and our notes on EU sensory profiling methods and US sensory methodologies compare how panels are run in each market.

For brands & formulators

If you are launching a new flavour in the US or EU, FRL’s sensory research and analytics team designs the validation study: the brief, attributes, panel and consumer plan, sample size and decision rule. We run or arrange the trained-panel, consumer and shelf-life sensory testing through accredited partner laboratories and local fieldwork partners, and report the results against the criteria you agreed.

If the flavour still needs work, our new food product development team formulates and iterates the recipe, and our regulatory and scientific research team reviews label and claim wording for each market. To start, contact our team with your concept, target markets and benchmarks.

In this series

• Flavour trends 2026: swicy, botanicals and global heat in the US and EU.
• Natural flavour labelling: FDA vs EU: how each market defines and labels natural flavourings.
• Off-odour and taint food recalls: how sensory defects become withdrawals and recalls.
• Designing a sensory study: the pillar guide to tests, panels and statistics.
• Flavour validation with sensory panels: a worked example for a sweet-heat seasoning. You are here.
• Sensory testing regulations in the US and EU: ten questions on rules, standards and claims.
• Service: Sensory research and analytics at Food Research Lab.

1. International Organization for Standardization. (2016). ISO 13299:2016 Sensory analysis: Methodology: General guidance for establishing a sensory profile. ISO. https://www.iso.org/standard/58042.html (cited for the descriptive profile method: vocabulary, scaling, replication and statistical analysis).
2. International Organization for Standardization. (2023). ISO 8586:2023 Sensory analysis: Selection and training of sensory assessors. ISO. https://www.iso.org/standard/76667.html (cited for screening, selecting and training assessors).
3. International Organization for Standardization. (2014). ISO 11136:2014 Sensory analysis: Methodology: General guidance for conducting hedonic tests with consumers in a controlled area (amended 2020). ISO. https://www.iso.org/standard/50125.html (cited for central location hedonic tests, presentation order and consumer recruitment).
4. Cvetan, D. (2025, September 29). Outlook 2026: Flavor trends. Food Technology, Institute of Food Technologists. https://www.ift.org/food-technology-magazine/outlook-2026-flavor-trends (cited for the continuing “swicy” sweet-and-spicy trend, including hot honey).
5. McEvoy, E. (2026, February 6). 2026 flavors to watch. Nutritional Outlook. https://www.nutritionaloutlook.com/view/2026-flavors-to-watch (cited for swicy, with hot honey and spicy mango, reported from Kerry’s 2026 Global Taste Charts).
6. Lawless, H. T., & Heymann, H. (2010). Sensory evaluation of food: Principles and practices (2nd ed.). Springer. https://doi.org/10.1007/978-1-4419-6488-5 (cited for carry-over and adaptation control, consumer test design and preference mapping).
7. Julious, S. A. (2004). Sample sizes for clinical trials with Normal data. Statistics in Medicine, 23(12), 1921–1986. https://doi.org/10.1002/sim.1783 (cited for the sample-size formula for non-inferiority comparisons of means).
8. International Organization for Standardization. (2021). ISO 4120:2021 Sensory analysis: Methodology: Triangle test. ISO. https://www.iso.org/standard/76666.html (cited for the triangle test, its difference and similarity versions, and its critical-value tables).
9. International Organization for Standardization. (2021). ISO 20784:2021 Sensory analysis: Guidance on substantiation for sensory and consumer product claims. ISO. https://www.iso.org/standard/69080.html (cited for the evidence needed behind taste and preference claims).