Smart Product Test: testing several recipes quickly, on a small budget, without sacrificing depth

Doing more with less: that is the equation many R&D and marketing teams face today. Less product available, less time, less budget, yet the same need for sound decisions. The Smart Product Test, developed by Repères, was designed precisely for these situations: an agile protocol for comparing several formulas on a small sample (e.g. 60 units per product), with first results 2 days after fieldwork, without sacrificing diagnostic depth.

The starting point: several recipes, little room for manoeuvre

The situation is a familiar one. Your R&D team has developed several formulas for an innovation or a renovation, or you are hesitating between several variants (fragrances, flavours, textures) for a range. You need to identify the recipe(s) with the strongest potential, possibly against a benchmark, and understand their strengths and weaknesses to guide optimisation.

But the constraints keep piling up:

• a limited quantity of product, often at prototype stage;
• tight deadlines, because the development schedule won't wait;
• a tight budget, especially when many options need testing;
• a need for rich, actionable insights, not just a ranking.

Conventional protocols usually force you to give up one of these imperatives.
The Smart Product Test was designed to deliver on all of them at once.

A streamlined yet discriminating protocol

The Smart Product Test is built on five complementary design choices.

A sequential monadic protocol. Each participant evaluates up to 5 products in a single session. Beyond that, a balanced incomplete block design makes it possible to evaluate more recipes without increasing the workload for each respondent.

• A central location test. The CLT format requires less product and is quicker to run than in-home fieldwork. An HUT remains possible, on the same principles.
• A small sample. 60 respondents are enough, so the quantity of product required is reduced (60 units per recipe).
• Several complementary modules. Products are evaluated from different perspectives, both rational and emotional, to maximise the test's ability to bring out differences.
• A hybrid qual-quant approach. Short qualitative interviews can extend the quantitative phase with selected participants for deeper insight.

Several levers to tell close products apart

With a small sample, conventional liking scales are rarely enough to separate similar recipes. Two products can show the same liking score while consumers clearly perceive differences. The Smart Product Test questionnaire therefore combines traditional measures with more intuitive modules that capture both System 1 (fast, emotional reactions) and System 2 (considered judgement).

1. R3mScore: measuring emotional engagement in three words

The first question asked: “What are the first three words that spontaneously come to mind?”. With no sentence or justification required, the answer stays as close as possible to the automatic reaction. A proprietary algorithm, based on the dimensional theory of emotions, weights each word (part of speech, rank, valence) to calculate a positive or negative emotional activation score. The words themselves reveal the functional and emotional drivers of the product experience. The tool draws on 10 years of use and over 6,000 stimuli scored in more than 20 countries.

2. The Alpha score: a single KPI for decision-making

The Alpha score combines stated overall liking and the R3mScore in equal parts, on a scale from 0 to 100. Built on more than 129,000 evaluations (17 studies, 9 categories, 12 countries), it improves discriminating power by 20% on average compared with overall liking alone. Calculated at individual level, it lends itself to any analysis by segment, country or wave, and comes with directly actionable performance thresholds, from rejection to love at first sight.

3. Open-ended questions with AI follow-up probing

Open-ended questions are often the weakest part of a self-administered questionnaire. Here, AI reads each answer and generates a personalised follow-up question in real time. A respondent who mentions “the texture” as an area for improvement is asked to specify what bothers them; they then explain that it is too runny and that they would like it creamier and easier to eat with a spoon. Precise, detailed answers that enrich the learnings and recommendations.

4. The CATA module: bringing out perceptual rough edges

CATA (Check All That Apply) presents a word wall on which respondents click to describe the product: smooth, too sweet, bland, natural… Quick and intuitive, it identifies the salient attributes of each recipe, its strengths and weaknesses, whether specific to it or shared with the other stimuli, and reveals differences that agreement scales struggled to bring out.

5. Optional: immediate qualitative insight

At the end of each quantitative session, one participant is selected, based on their answers and a criterion defined beforehand, for an interview of around 20 minutes. These discussions make it possible to explore an optimisation route in greater depth, compare a prototype with the current offer or a competitor, or better understand the sensory and emotional expectations associated with the category.

Results delivered very quickly

The protocol's agility carries through to reporting, shared progressively with project teams:

D+2: dashboard of the main results on our online platform

D+4: first results in Excel, closed questions and verbatims.

D+9: report in Excel with summary, analysis of open-ended questions, penalty analysis, etc.

D+10: qualitative report (if the option is selected).

Automated formatting of results, together with the reduced sample size, helps optimise costs, even when a large number of products is tested.

The Smart Product Test meets a specific need: screening several formulas under product, time and budget constraints. Its strength lies in combining a streamlined protocol with modules that multiply its discriminating power: R3mScore, CATA, Alpha score and AI-probed open-ended questions.
Less product, faster results, controlled costs, yet still a fine-grained reading of what makes the difference between two recipes.
The methodology can be deployed in France and internationally.

👉 Want to know more? Contact us!