Sensory evaluation panels

01Overview and value chain#

Markers EC: Food information to consumers & sensory-claim substantiation | OECD: Bioeconomy policy & governance | Regulator: FDA (US), EFSA (EU)

Sensory evaluation panels are specialist service firms that recruit, screen and train panelists to formally assess taste, aroma, texture and appearance, then translate what they taste into quantitative data manufacturers use for reformulation, quality control and label-claim substantiation. MMR Research Worldwide runs an FMCG-focused sensory-science practice combining trained descriptive panels with “sensory AI drivers” that model how measured attributes predict consumer liking, and in 2026 published its first peer-reviewed paper in the Journal of Food Quality and Preference. Sensory Spectrum, a US descriptive-analysis specialist for over two decades, maintains the proprietary Spectrum Method and physical intensity references — including a 7.5-point Universal Aromatic Intensity Scale anchor recalibrated in 2026 — and runs a long-running Descriptive Panel Leadership training course. AZTI, the Spanish marine and food-technology research institute, operates an expert-panelist analysis service alongside consumer sensory studies (acceptability tests, home-use tests, benchmarking). Merieux NutriSciences runs a dedicated sensory and consumer-research practice offering sensory claim validation and reformulation-impact testing across food categories. Davis Sensory Institute trains panel leaders in discrimination testing methodology, including Thurstonian modeling for difference testing, and partners with specialist education providers on continuing-education panel-leader certification. Intertek operates sensory-evaluation services as part of its food-testing network, spanning product-success testing for flavor houses and expanded Greater China food-safety and sensory testing capacity added in 2026.

The key directions of sensory evaluation panels are:

1. Descriptive panel training and calibration: recruiting and training panelists on a defined attribute vocabulary and intensity scale, then calibrating them against physical references so scores are reproducible across sessions and panelists.
2. Discrimination and difference testing: running triangle, duo-trio or Thurstonian-model difference tests to determine whether a reformulated product is perceptibly different from the original before a manufacturer commits to a full launch.
3. Consumer acceptability and claim-substantiation testing: running larger, untrained consumer panels (central-location tests, home-use tests) to validate liking scores and marketing claims such as “improved taste” against a documented, defensible protocol.
4. AI-assisted sensory-to-liking modeling: building predictive models that link trained-panel attribute scores to consumer liking data, letting a manufacturer estimate market response to a reformulation before running a full consumer study.

Sectoral value chain#

[Panelist recruitment & screening] ──> [Descriptive vocabulary & scale calibration] ──> [Trained panel sessions]
                                                                                                 │
                                                                                  (Discrimination/difference testing: triangle, Thurstonian)
                                                                                                 │
     [Reformulation / claim decision] <──── [Sensory-to-liking data modeling] <─── [Consumer acceptability testing]

Value chain levels#

Cross-cutting technologies of the sector:

• Descriptive sensory analysis: a structured methodology (such as the Spectrum Method) in which trained panelists score defined attributes against calibrated physical intensity references, producing reproducible quantitative data rather than a simple preference verdict.
• Thurstonian discrimination testing: a statistical model for difference tests (triangle, duo-trio) that estimates the true perceptual distance between two products from panelist response patterns, giving a more precise discrimination measure than a simple pass/fail count.
• Sensory-AI drivers: predictive models that link trained-panel attribute-intensity scores to consumer liking data, letting a manufacturer forecast market response to a reformulation before commissioning a full consumer study.

02US#

The United States hosts the most established descriptive-panel and discrimination-testing capacity, anchored by long-running specialist firms serving CPG and flavor-house clients.

Sensory Spectrum’s proprietary Spectrum Method, Davis Sensory Institute’s Thurstonian-model panel-leader training, Merieux NutriSciences’ sensory-claim validation#

• Sensory Spectrum: maintains the proprietary Spectrum Method and physical intensity references, including a 7.5-point Universal Aromatic Intensity Scale anchor recalibrated in 2026, and runs a long-running Descriptive Panel Leadership training course for panel leaders across the industry.
• Davis Sensory Institute: trains panel leaders in discrimination-testing methodology, marking close to a century of Thurstonian modeling in 2026 and partnering with specialist continuing-education providers on panel-leader certification.
• Merieux NutriSciences: runs a dedicated sensory and consumer-research practice offering sensory claim validation and reformulation-impact testing, positioned around ingredient-change impact on taste, texture and aroma for repeat-purchase risk.

03CN#

China’s sensory-testing capacity runs through general-purpose third-party food-testing labs and industry conference circuits rather than a dedicated specialist brokerage at the scale of the US or EU firms profiled here.

Third-party lab-based sensory testing, an active conference and standards circuit, Intertek’s expanded Greater China testing capacity#

• General third-party testing labs: Chinese food-safety testing platforms (foodmate.net-affiliated labs, regional testing centers) offer sensory-evaluation services as one line item within broader food-safety and quality testing, rather than as a dedicated descriptive-panel specialty comparable to Sensory Spectrum or MMR.
• Active conference circuit: an industry sensory-evaluation practice-application seminar (Shanghai) and related food-technology-innovation forums recur through 2026, indicating a growing but still-consolidating professional community rather than a small number of dominant vendors.
• Intertek Greater China: expanded its Greater China food-safety testing capacity in 2026 (including cereulide testing), providing an international-standard testing option operating alongside domestic third-party labs, though no China-based sensory-panel specialist matching the scale of the US/EU firms could be independently confirmed via a live 2026 source.

04EU#

The European Union’s sensory-testing capacity is anchored by applied research institutes running both expert and consumer panels, serving food and marine-product manufacturers.

AZTI’s expert-panelist and consumer sensory-study service, MMR Research’s peer-reviewed sensory-AI methodology#

• AZTI: the Spanish marine and food-technology research institute operates an expert-panelist analysis service alongside consumer sensory studies — acceptability tests, home-use tests and benchmarking — serving food and seafood manufacturers.
• MMR Research Worldwide: UK-headquartered, runs an FMCG-focused sensory-science practice combining trained descriptive panels with sensory-AI-driver modeling, and published its first peer-reviewed paper in the Journal of Food Quality and Preference in 2026.
• Intertek: operates sensory-evaluation services within its broader European testing network, supporting product-success testing for flavor houses and CPG manufacturers alongside its global testing infrastructure.

05Leading companies and research institutes#

06Tech stack and innovations#

The sensory-evaluation stack combines calibrated human panels with statistical discrimination modeling and, increasingly, predictive analytics:

1. Descriptive sensory analysis (Spectrum Method and equivalents):
   • Trained panelists score defined attributes against calibrated physical intensity references — such as a specific citrus-juice product serving as a fixed 7.5-point aroma-intensity anchor — producing reproducible quantitative data a manufacturer can track across reformulation cycles.
2. Thurstonian discrimination testing:
   • Triangle and duo-trio difference tests are analyzed with a Thurstonian statistical model estimating the true perceptual distance between two products from panelist response patterns, giving a more precise discrimination measure than a raw correct-guess tally.
3. Sensory-AI drivers:
   • Predictive models link trained-panel attribute-intensity scores to consumer liking data, letting a manufacturer forecast market response to a reformulation option before commissioning a full, more expensive consumer study.

07Value chains and production pipelines#

Industrial pipeline for validating a food reformulation through a sensory-panel study#

┌───────────────────────────┐      ┌───────────────────────────┐
│ 1. Panelist recruitment &  │ ───> │ 2. Vocabulary & scale      │
│    screening                    │      │    calibration                   │
└───────────────────────────┘      └───────────────────────────┘
                                                 │
                                                 ▼
┌───────────────────────────┐      ┌───────────────────────────┐
│ 4. Discrimination/          │ <─── │ 3. Trained descriptive     │
│    difference testing           │      │    panel session                │
└───────────────────────────┘      └───────────────────────────┘
              │
              ▼
┌───────────────────────────┐      ┌───────────────────────────┐
│ 5. Consumer acceptability   │ ───> │ 6. Sensory-to-liking data   │
│    testing                      │      │    modeling & report            │
└───────────────────────────┘      └───────────────────────────┘

Stage 1: Panelist recruitment and screening

The panel provider screens candidates for taste and aroma acuity and availability, building a standing roster of trained panelists for a given product category.

Stage 2: Vocabulary and scale calibration

Panelists are trained on a shared attribute vocabulary and calibrated against physical intensity references, such as a fixed aromatic-intensity anchor product, so scores stay reproducible across sessions.

Stage 3: Trained descriptive panel session

The calibrated panel scores the test product against the reformulated or competitive samples on the defined attribute scale, producing a quantitative intensity profile.

Stage 4: Discrimination and difference testing

Where the question is whether two samples are perceptibly different at all, a triangle or duo-trio test is run and analyzed with a Thurstonian model to estimate the statistical significance of any perceived difference.

Stage 5: Consumer acceptability testing

A larger untrained consumer panel — via central-location testing or home-use testing — scores liking and purchase intent, providing the data needed to substantiate a marketing claim.

Stage 6: Sensory-to-liking data modeling and report

The provider models the relationship between trained-panel attribute scores and consumer liking outcomes, delivering a report the manufacturer uses to finalize the reformulation or claim decision.