Sensory Tasting in the Winery

August 31, 2020

Erin Norton

Part 3: Interpreting Sensory Test Results

In this series of articles, Erin Norton (Education & Outreach Coordinator) will provide details on how you can use sensory evaluation in your winery.  Any questions or comments, please direct them to Erin at

At this point in the process, you have completed your sensory test, and have a bunch of data to compile and interpret.  If you are interested to know what a sensory test can tell you, or how to set one up in your winery, please see the previous two newsletters, with those topics (April 2020-When to use a sensory test in your winery; June 2020-How to set up a sensory test in your winery). Also, I’d like to encourage you to reach out to staff members at the MGWII if you have questions regarding setting up any sensory test, even down to setting up spreadsheets, what supplies you’ll need, or what questions are the right questions to ask.  We have performed sensory evaluation studies numerous times and are familiar with setup and execution of these studies. 

Back to data interpretation.  The first thing that must be done, no matter what type of sensory test you used, is to compile and input your data into a software program like Excel.  Instead of counting any results and pooling them before entering into a spreadsheet, input the raw data itself.  This will let you manipulate your raw data directly. For example, if you had people rate your wine on a 5-point scale, do not count all of the votes for Extremely Like and only input that into your spreadsheet, instead have each vote as a separate line in the spreadsheet. (See the Figure Below)

Once you have all your raw data input, you can start to manipulate the data to understand the answer to your questions.  In addition, this brings me to the point of “What can the data tell you?”  The only solid conclusion you can make is based on the actual question that you asked your panelists.  Therefore, for the above example of “liking”, you can only answer how much the participants liked that wine.  If you asked them to rate multiple wines, with this question above, you cannot compare the wines based on their overall scores.  You would have needed to ask the participants which wine they preferred between the two. 

Sticking with the example above.  You can do several things.  You could tally all of the votes in each column and create a bar graph, or you could give each column a score or weight and tally the votes.  The two interpretations would be represented in the graphs below.  

As you can see, you will produce different pictures.  In this case I would choose to use the Sum-graph,  but in certain cases a Weighted Sum may be appropriate.  The weighted sum can give you an overall average liking of the wine (in this case it is 3.2/5). 

Another consideration is to use demographic data (if you collect it in your study).  For this situation I would recommend using ranges and percentages/averages.  The data can be sorted into age (this can be easily done in Excel by highlighting the age column, selecting Sort in the far right of the Home menu, be sure to expand the selection to have the other columns rearrange along with the age column).  You could also sort by M/F.  For these two situations I would present the data as an average over a given data range. 

The above data manipulations are most useful for consumer type of sensory tests where you ask a large number of consumers a straightforward question like rating or yes/no type of questions. 

For sensory tests where trained panelists are used, and more targeted questions are asked, data manipulation can become more complex.  If some questions asked of panelists have simple yes/no or rating answers, these can be treated like a larger consumer sensory test.  Remember that there will be less data points with a trained group of participants (since it is common to only use around 10 participants), however the data should be more reliable since these people are trained in wine tasting, and will hopefully provide results that are generally in agreement. 

If panelists were asked to score attributes, for example the intensity of acidity and sweetness of a particular wine on a line scale (Figure below), then scores are measured on the line in length (typically in centimeters).  These scores can be averaged and left as a number out of 15 (if that is the total length of your line), or divided by 15 to get a number that is independent of the length of the line.  In the example below the score for acidity is ~5.2cm or 0.35, and the score for sweetness is ~9.5cm or 0.63.  The corresponding scores for all the panelists can be averaged to get an average score for that attribute for that wine.

Where things become more complicated is if you want to see if certain attribute scores or demographics are correlated.  For example, do all panelists rate the wines similarly, or do male participants rate attributes differently to female participants (you want every panelist to rate the wine similarly for the data to be robust).  If panelists are indeed in harmony with their scoring, then you can decide if wines are similar or not.  If your question was concerning the use of a certain yeast, and you hoped that the new yeast produced more fruity aromas, then you would compare all the results for fruit aroma between wines.  Hopefully it would show you that they are indeed different.  How do you calculate that?  You need statistical software.  This is where the institute can help you analyze your data and draw the correct conclusions from your study.  At this point, I will reiterate that it is important to clearly set up your sensory study with the proper questions so that the data can be used to draw conclusions.

Therefore, consumer type sensory studies are more straightforward to analyze because the questions themselves are more simple.  Typically, averages, or percentages are calculated and used to describe this type of data.  Excel is a good program to use to set up some of these calculations, especially if you have up to 100 participants.  Trained panel sensory studies can have more complicated calculations and analysis since the questions are more specific and require trained wine tasters to answer them.  Please feel free to reach out to the institute ( to analyze these types of data sets.

This concludes the series for Sensory Testing in the Winery.  I hope it has inspired some of you to try gathering  sensory evaluation data yourself to see what wines consumers like, or to tell if wines are different using different treatments.  You may not be ready to set up your sensory study right away, but the institute is always available to guide you through the process when you reach that point. 

If you are new to Excel, here is a short YouTube video to show you how to use some core functions: 

Some additional reading if you are interested in Sensory Evaluation: