Taste Testing That Tells R&D Exactly What to Fix and By How Much

Most taste tests produce a preference ranking. SIS International produces a reformulation roadmap. We run blind taste tests, triangle tests, and sensory panels at our Flatiron District test kitchen with temperature-controlled storage, standardized serving protocols, and controlled lighting. JAR scales and penalty analysis isolate the specific attributes dragging down overall liking and quantify the impact of each. R&D gets a prioritized fix list, not a thumbs-up or thumbs-down.

Taste testing: blind sensory panel evaluation at SIS Flatiron District test kitchen

Blind Taste Tests with Sequential Monadic Design

Respondents evaluate one product at a time in randomized order, eliminating the comparison bias that side-by-side testing introduces. Each sample is coded, served at controlled temperature, and evaluated on 9-point hedonic scales for overall liking plus JAR (Just About Right) scales for specific attributes: sweetness, saltiness, bitterness, carbonation, texture, mouthfeel, and aftertaste. We recruit 150-500 respondents per CLT by category consumption frequency, flavor preference, and demographic profile. Recent projects: sparkling water flavor ranking, plant-based burger texture optimization, and RTD cocktail sweetness calibration.

Triangle Tests for Reformulation Validation

When a manufacturer changes an ingredient, adjusts a supplier, or modifies a process step, the triangle test determines whether consumers can detect the difference. Three coded samples are presented, two identical and one different, and the respondent identifies the odd one out. A statistically significant detection rate means the reformulation is perceptible. A non-significant result means R&D can proceed without consumer-facing risk. Coca-Cola and PepsiCo both run triangle tests before any formulation change reaches production because a detectable shift in a flagship product can trigger consumer backlash at scale.

Penalty Analysis and JAR Diagnostics

Overall liking scores tell you whether consumers prefer Product A or Product B. They do not tell you why, or what to change. Penalty analysis takes JAR (Just About Right) data and calculates the mean drop in overall liking caused by each off-target attribute. If “too sweet” reduces liking by 1.8 points on the 9-point scale while “not salty enough” reduces it by 0.4 points, R&D knows to prioritize sweetness reduction. SIS delivers a penalty matrix that translates sensory data into a ranked reformulation action list with quantified impact estimates.

Paired Comparison and Protomonadic Testing

Paired comparison presents two products and asks which is preferred. Protomonadic design evaluates each product independently first (monadic), then presents both for a direct comparison. The protomonadic approach captures both absolute performance and relative preference, revealing cases where a product scores well on its own but loses in head-to-head competition. SIS uses protomonadic designs for competitive benchmarking studies where the client needs to understand how their product performs both in isolation and against the category leader.

Concept-Product Fit Testing

A product concept that scores well in concept testing can still fail if the formulation does not deliver on the promise. SIS runs concept-product fit studies where respondents first evaluate the concept (packaging, claims, positioning), then taste the product and report whether it met, exceeded, or fell short of expectations. The gap between concept appeal and product delivery identifies where marketing overpromised or where the formulation underdelivered. A functional beverage client discovered that their “smooth energy” concept generated strong purchase intent, but the actual product’s bitter aftertaste from the caffeine source contradicted the “smooth” positioning.

Alcohol, Beer, and Spirits Evaluation

Alcohol sensory testing requires specific protocols: controlled serving temperature by spirit category, standardized glassware, palate cleanser intervals, and responsible consumption limits that cap the number of samples per session. SIS recruits by drinking occasion (on-premise vs. off-premise), spirit category affinity, and consumption frequency. We evaluate aroma, initial palate, finish, mouthfeel, and overall balance. A craft brewery used our sensory panels to identify that their flagship IPA’s hop bitterness was 15 IBU above the preference threshold for their target demographic, leading to a recipe adjustment that improved repeat purchase rates.

SENSORY RESEARCH INTELLIGENCE

Three Testing Errors That Send the Wrong Product to Market

DIAGNOSIS 01 //

SIDE-BY-SIDE COMPARISON BIAS

The Paired Test Distortion

Side-by-side taste tests exaggerate small differences. Respondents who cannot distinguish two products in isolation detect a “preference” when presented simultaneously because the testing format forces a choice. A snack brand reformulated based on paired comparison data showing a strong preference for the new recipe. In-market, where consumers eat the product alone, the preference disappeared. Sequential monadic design evaluates each product independently in randomized order, producing liking scores that reflect how consumers experience the product in real life.

DIAGNOSIS 02 //

PREFERENCE DATA WITHOUT DIAGNOSTIC DATA

The “Which One Wins” Dead End

A taste test shows 58% preference for Variant B over Variant A. The product team selects Variant B. But R&D does not know what drove the preference, what the losing variant’s fixable weakness was, or whether a third formulation combining attributes from both would outscore either. Without JAR scales and penalty analysis, the test produces a winner but no understanding of why it won. The best formulation may not have been in the test. SIS designs every taste test to produce both a preference outcome and a diagnostic roadmap.

DIAGNOSIS 03 //

UNCONTROLLED TESTING ENVIRONMENT

The Conference Room Taste Test

Internal taste tests in conference rooms introduce uncontrolled variables: inconsistent product temperature, ambient food odors, visual cues from packaging, and social influence from colleagues tasting simultaneously. Coca-Cola runs every sensory evaluation in a controlled environment with standardized lighting, temperature, and serving protocol for a reason. SIS operates a dedicated test kitchen in Manhattan’s Flatiron District with temperature-controlled product storage, neutral lighting, individual tasting stations, and standardized serving ware. The environment is the control. Without it, the data is noise.

PRIMARY RESEARCH DELIVERABLES

What SIS Delivers to Food and Beverage R&D Teams

Controlled CLT with Penalty Analysis and Reformulation Roadmap

150-500 respondents per study at our Flatiron District facility. Sequential monadic design with 9-point hedonic scales and JAR attribute measurement. Penalty matrix quantifying the liking impact of each off-target attribute. The deliverable is a ranked reformulation priority list R&D can act on directly.

Triangle Tests for Ingredient and Process Changes

Statistically powered triangle tests that determine whether consumers can detect a reformulation, supplier change, or process modification. The output is a go/no-go recommendation with detection rate and confidence interval. If the change is imperceptible, production proceeds without consumer risk.

Competitive Benchmarking with Protomonadic Design

Your product evaluated independently and head-to-head against named competitors. Absolute performance scores plus relative preference data. Attribute-level diagnostics identify where you lead and where the competitor outperforms, with specific sensory dimensions attached to each gap.

Concept-Product Fit Validation

Sequential evaluation of concept appeal followed by product tasting with expectation-versus-delivery gap measurement. Identifies where the formulation falls short of the marketing promise and where it exceeds it. The output tells both R&D and marketing which claims the product can credibly support and which need adjustment.

THE SIS DIFFERENCE

تشمل عروض خدماتنا ما يلي: