Reb A vs Reb M: Why Real-Base Sensory Testing Matters

Screen the ingredient, then test the intended product

The illustration separates a neat-ingredient water screen from testing the complete product, where the rest of the formula remains present. Its 90-second note is an example checkpoint to agree before a trial, not the timing used in the published study below. The scene does not show measured factory results or a winning sample.

A water screen is only the first filter

An equal-sweetness water test can help remove an obviously unsuitable sample. It cannot show how a sweetener will interact with milk protein, soy protein, fat, acid, aroma, minerals, viscosity or the finished process. Keep the water screen, but do not treat it as finished-product approval. If the finished product is a complete powder meant to be mixed with water, that full powder at the intended dilution is the real base. Do not remove its protein, flavor or other ingredients and call a neat-sweetener solution the same test.

Name the matrix before the sample

Write down the product before deciding which sample is better: clear beverage, protein drink, soy beverage, milk drink, yogurt drink, coffee, tea or another base. Add pH or acidity, soluble solids, protein or fat level, flavor, serving temperature, heat treatment, mixing order and storage plan. Kim and Hong's 2023 milk/soymilk study found flavor-dependent sensory effects; its chocolate and vanilla results do not establish an improvement for every flavor or formula.

• Kim and Hong (2023): published milk/soymilk study abstract

Compare sweetness and bitterness over time

The first sip and the remaining taste are separate observations. Record onset, peak, lingering sweetness, bitterness, astringency, aroma and mouthfeel instead of reducing everything to one preference score. Agree the assessment times, whether the sample is swallowed or expectorated, the serving amount and palate-cleansing procedure before the comparison.

• Tao and Cho (2020), Foods 9, 1026: original study and Table 3

Table values are means ± SEM from Table 3, on the study's 15-cm line scale; 5 s and 60 s follow expectoration. The paper used specific 95% glycoside products in water and also included Reb D, which is not plotted here. Do not convert these intensity results into consumer liking, a universal substitution ratio or an equivalence claim for our supplied grades.

Keep the control and comparison question identifiable

A useful first round can include the current or sucrose control, Reb A, Reb M and a blend only when the blend answers a real formulation question. Choose whether the trial compares the same addition or the amount needed to reach an agreed sweetness, then record the actual addition and its unit for every sample. Those two questions should not be mixed in a single ranking. Keep serving conditions, flavor and process consistent, or record the change explicitly.

Check the finished process

For a beverage, record dissolution, haze, sediment, separation, carbonation if relevant, heat treatment, filling and storage. For a dairy or protein base, record mixing order, foam, viscosity, protein note and cold storage. A sample that tastes clean before processing may need a second check after the real process and storage window. Keep physical observations separate from taste: a haze or sediment note is not a chemical solubility measurement, and a clean first sip does not establish storage stability.

Use the existing blank sensory record

The free Excel record already supports one coded sample, panelist and checkpoint, with six sensory/preparation dimensions and separate ingredient, safety and document gates. Add actual timed observations in its observation field and keep the study numbers above separate from your own trial. Its optional 1–5 scores describe fit to your agreed target; they are not the study's 15-cm intensity ratings and must not be converted between scales. Leave untested items blank or describe why they do not apply. No email is needed, and there is no automatic total, ranking or approval.

• Download the blank sensory record (.xlsx, no email)
• How to use target-fit scores and separate decision gates

Use market rules as a boundary

JECFA's 2026 Chemical 267 entry is an official evaluation source for steviol glycosides and purified Reb M, including the current ADI expressed as steviol. It is not a blanket approval for every commercial grade or every market. Confirm the target country, food category, glycoside identity, production route, label name and document set before a launch or quotation is treated as final.

• View stevia specifications

Turn observations into the next sample question

Start with the product, market and the taste or preparation issue you want to resolve; share what you already know. A useful feedback note names the sample and base, the comparison control, the actual condition and what happened at the agreed checkpoint. Add packaging, estimated quantity, required documents and sample needs as the project develops. You can contact us before the full brief is complete.

• Match the sample to the grade and purchasing brief

Sources and their scope

Tao and Cho (2020), Foods 9, 1026, supplies the water-study data; Kim and Hong (2023), Journal of the Science of Food and Agriculture, 103, 7445–7454, supports flavor/matrix-specific testing. The WHO/FAO JECFA Chemical 267 entry includes the 2026 evaluation. These are research and evaluation references, not company trial results or approval of a particular finished product.

• WHO/FAO JECFA: steviol glycosides, Chemical 267

Common questions