Skip to content
Sensory analysis / Sensory profiling

What is inside your product, attribute by attribute

The sensory profile is the quantitative, objective description of a product: appearance, odour, body and flavour, each scored on a scale that means the same in any category and in any year.

The literature calls it quantitative descriptive analysis, or QDA. It describes; it does not judge whether the product is liked.

Descriptive profile · trained panel

Each axis is an attribute; the dashed line is the category reference

SweetnessAcidityCreaminessFruity aromaPersistence

Illustrative chart. The values do not belong to any real product.

Four steps, and the fourth decides

A descriptive profile without statistical validation is a table of numbers that looks like data. These are the steps, and the order matters.

  1. The vocabulary is built

    The panel generates and agrees the attributes that describe the category: appearance, odour, body and flavour. There is no closed list to start from, because every category has its own.

  2. The scale is anchored

    Every attribute is referenced against known standards, which is why the score is absolute and not relative to whichever samples are on the table that day. How that scale is trained is explained in SensoExpert Lab.

    The panel and its scale →
  3. Assessment runs in replicates

    Samples are served coded and in balanced order, with replicates. Without replication there is no way to know whether a difference belongs to the product or to the day.

  4. It is validated before it is read

    We check that the panellists discriminate, repeat and agree with each other. A profile that fails that check is not delivered: the session is repeated.

A table, a chart and a reading

That is what you receive once the profile has passed validation. The panel produces the first two; the third is what turns the numbers into a decision.

The table
Every attribute with its score and its dispersion across panellists, for all the references measured.
The chart
Your product’s profile drawn against the references we have measured, attribute by attribute.
The reading
Which differences are real, which fall within the noise and which the consumer will perceive. That last distinction is the one most often skipped and the most expensive to skip: a statistically significant difference is not always a difference anyone notices when eating.

The profile describes the product. It does not say whether it will be liked, or by whom.

For that you take it to the consumer and cross the two measures. And if you only need to know whether a change shows, there is a shorter test.

Which product do you want to describe?

Tell us the category and how many references need measuring. We tell you which attributes that category has and what you will be able to claim with the profile in hand.