Skip to main content

Wine online and digital decanting: Naturalness effects in large-scale consumer reviews of wine

Wine Economics and Policy, 2026

Michał Folwarczny, Magalie Dubois, Vincent Clerc, Bruno Schivinski, Tobias Otterbring

Wine Economics and Policy, 2026, just accepted

  • Open access
  • 6.4 million Vivino reviews
  • 10,614 wines
  • Computational linguistics
  • Sentiment analysis
  • Natural wine

Free full text: the open access accepted manuscript (CC BY 4.0, just accepted), re-typeset as a clean PDF. It has not yet been copyedited or typeset by the journal; please cite the published version.

Keywords
  • natural wine
  • naturalness bias
  • psychological essentialism
  • authenticity
  • sentiment analysis
  • computational linguistics
  • electronic word-of-mouth
  • online wine reviews
  • Vivino
  • consumer ratings
  • price premium
  • credence attributes
  • information asymmetry
  • expectation disconfirmation
  • natural-is-better bias
  • minimal-intervention wine
  • discrete emotions
  • disgust
  • review length
  • text mining
  • AFINN lexicon
  • NRC Emotion Lexicon
  • negative binomial regression
  • R (tidytext)

Natural wines cost about 20% more and get longer reviews than conventional wines, but consumers rate them lower and write about them with less positive and more disgusted, angry language. In 6.4 million English-language Vivino reviews, naturalness worked as a quality signal that raised expectations the wine often did not meet.

Key findings

  1. Natural wines carry a price premium of about 20%. Controlling for rating, vintage, alcohol, sensory characteristics and country of origin, wines that Vivino classifies as natural were priced about 20% higher than conventional wines (b = 0.18) (Folwarczny et al., 2026).
  2. Yet consumers rate natural wines slightly lower. Natural wines received lower ratings (M = 4.09 vs 4.12 on a 5-point scale), and the gap held with controls (b = −0.02, p < .001), contrary to the natural-is-better hypothesis (Folwarczny et al., 2026).
  3. Natural wines get longer but less positive reviews. Reviews of natural wines were about 20% longer (23.0 vs 19.2 words; incidence rate ratio = 1.20) but had lower per-word sentiment (2.05 vs 2.21 on the AFINN scale) (Folwarczny et al., 2026).
  4. Disgust and anger rise most. Natural wine reviews contained 40% more disgust words, 32% more anger words and 30% more fear words than conventional wine reviews, with only small increases in joy (+2.5%) and surprise (+1.8%) (Folwarczny et al., 2026).

At a glance

Data
Consumer reviews scraped from Vivino (September–October 2025), covering the platform’s full history; 12.9 million observations, of which 6,395,585 English-language reviews with a natural-wine classification were analysed
Wines
697 natural and 9,917 conventional wines, using Vivino’s own ‘is_natural’ attribute; 116,747 reviews of natural wines
Outcomes
Rating (1–5), price (USD, log), review length (words), per-word sentiment (AFINN) and discrete emotion word counts (NRC Emotion Lexicon)
Analysis
Independent-samples t-tests, then multiple linear and negative binomial regressions with controls (price or rating, vintage year, alcohol, acidity, body, tannin; country of origin for price), in R

Figure 1. Emotion word count differences between natural and conventional wine reviews

Negative emotions and sentimentPositive emotions and sentiment010%20%30%40%Disgust: +40.0% (p <.001)Disgust+40.0% ***Anger: +31.8% (p <.001)Anger+31.8% ***Negative (general): +31.3% (p <.001)Negative (general)+31.3% ***Fear: +30.4% (p <.001)Fear+30.4% ***Sadness: +7.5% (p <.001)Sadness+7.5% ***Positive (general): +5.8% (p <.001)Positive (general)+5.8% ***Joy: +2.5% (p <.001)Joy+2.5% ***Surprise: +1.8% (p <.001)Surprise+1.8% ***Trust: +0.6% (p .145)Trust+0.6% n.s.Anticipation: -2.7% (p <.001)Anticipation−2.7% ***Change in emotion word count, natural relative to conventional wine reviewsNRC Emotion Lexicon; values from Table 2 of the article; *** p < .001; n.s. = not significant (p = .145); 6,395,585 reviews.
Redrawn from the values in Table 2 of the article: percentage change in NRC emotion word counts for natural relative to conventional wine reviews. Negative emotions rise far more than positive ones; the difference in trust words is not significant.

Natural versus conventional wines: price, ratings and reviews

Market and engagementPrice premium+20%Review length+20%Emotion words in reviewsDisgust+40.0%Anger+31.8%Fear+30.4%Joy+2.5%
Percentage differences for natural relative to conventional wines, from the regression and t-test results reported in the article (6,395,585 reviews). Price premium and review length after controls; emotion words from Table 2.

Tables

Table 1. Descriptive statistics

Variable M SD Median Min Max Skewness Kurtosis
Rating 4.12 0.65 4.00 1.00 5.00 −1.07 2.67
Price (USD) 253.92 3,673.35 51.99 1.16 200,000 53.59 2,910.45
Review length (words) 19.23 18.64 12.00 1.00 112.00 1.79 3.05
Sentiment (per word) 1.68 1.39 2.00 −5.00 5.00 −0.56 −0.33
Alcohol by volume (%) 13.38 2.24 13.50 0.00 19.00 −5.00 27.08
Acidity 3.71 0.79 3.50 1.00 5.00 −0.65 0.21
Body 3.93 0.84 4.00 1.00 5.00 −0.63 0.02
Tannin 3.85 0.90 4.00 1.00 5.00 −0.72 −0.29
Sweetness 1.15 0.48 1.00 1.00 5.00 5.44 36.79
Fizziness 4.64 0.50 5.00 3.50 5.00 −0.75 −1.22
Intensity 4.03 0.88 4.00 1.00 5.00 −0.83 0.06
Vintage year 2013 14.03 2018 1806 2025 −3.13 20.83

N = 6,395,585 English-language reviews with valid natural wine classification. M = mean, SD = standard deviation. Sensory characteristics (acidity, body, tannin, sweetness, fizziness, intensity) measured on 1–5 scales.

Table 2. Emotion word counts in natural versus conventional wine reviews

Emotion Type M conv. SD conv. M nat. SD nat. Difference t p
Joy Positive Emotion 0.84 1.05 0.86 1.09 +2.5% −6.65 <.001
Trust Positive Emotion 0.79 1.04 0.80 1.06 +0.6% −1.46 .145
Anticipation Positive Emotion 0.68 0.95 0.66 0.95 −2.7% 6.54 <.001
Surprise Positive Emotion 0.43 0.71 0.44 0.73 +1.8% −3.63 <.001
Anger Negative Emotion 0.14 0.41 0.19 0.48 +31.8% −32.16 <.001
Disgust Negative Emotion 0.28 0.59 0.39 0.71 +40.0% −53.21 <.001
Fear Negative Emotion 0.12 0.38 0.16 0.44 +30.4% −28.32 <.001
Sadness Negative Emotion 0.23 0.53 0.25 0.55 +7.5% −10.72 <.001
Positive General Sentiment 1.94 2.01 2.06 2.05 +5.8% −18.65 <.001
Negative General Sentiment 0.45 0.79 0.59 0.92 +31.3% −51.68 <.001

Values are mean emotion word counts per review (NRC Emotion Lexicon). Percentage change = [(natural − conventional) / conventional] × 100. Conv. = conventional; nat. = natural.

The question

Consumers tend to see natural products as better, and natural wine is sold on what has not been done to the grapes. Because naturalness is a credence attribute that consumers cannot verify, it can shape both what they pay and how they judge the wine after drinking it. We asked whether naturalness changes consumers’ ratings, the prices of wines, and the length, sentiment and emotions of their online reviews.

Implications

Practical implications

Natural wine brands should position naturalness as a credence attribute, emphasising process history, rarity and authenticity rather than competing on quality-for-price, and manage expectations about variable sensory profiles, for example by presenting cloudiness or volatile acidity as signatures of minimal intervention. The long, emotional reviews also offer material for online brand communities.

Theoretical implications

The findings challenge the natural-is-better bias in a real market: naturalness raises valuation before purchase as a quality signal but lowers evaluation after consumption through expectation disconfirmation, so the price premium and the rating penalty are two stages of the same process.

Limitations and future research

  • Vivino users are more involved than the average wine buyer.
  • Natural status relies on Vivino’s own undisclosed classification and treats a heterogeneous category as one group.
  • Reviews are clustered within wines, so observations are not independent.

Questions this paper answers

Do people rate natural wine higher than conventional wine?

No. In 6.4 million English-language Vivino reviews, natural wines received slightly lower ratings than conventional wines, even after controlling for price and wine characteristics (Folwarczny et al., 2026).

How much more does natural wine cost?

About 20% more than comparable conventional wines on Vivino, after controlling for rating, vintage, alcohol, sensory characteristics and country of origin (Folwarczny et al., 2026).

What emotions do natural wine reviews express?

More negative emotions than conventional wine reviews, especially disgust (+40%), anger (+32%) and fear (+30%), with slightly more joy (+2.5%) and surprise (+1.8%) (Folwarczny et al., 2026).

Why would consumers pay more for a wine they rate lower?

Naturalness works as a signal of authenticity and process integrity before purchase, but it also raises expectations that the wine’s unconventional sensory profile may not meet after drinking (Folwarczny et al., 2026).

Key concepts

Natural wine
Wine made with minimal intervention, typically from organic or biodynamic grapes with indigenous yeasts and without additives such as sulfur dioxide; the term has no legal definition.
Naturalness bias
The tendency to see things of natural origin or made with little human intervention as better than conventional alternatives.
Credence attribute
A product quality, such as how a wine was made, that consumers cannot verify even after consumption.
Expectation disconfirmation
Dissatisfaction that arises when a product performs below the expectations formed before purchase.

Open science and materials

  • Full text (PDF) Open access accepted manuscript (just accepted), re-typeset as a clean PDF with the tables and figure
  • Published version Wine Economics and Policy (open access, just accepted)

How to cite

Folwarczny, M., Dubois, M., Clerc, V., Schivinski, B., & Otterbring, T. (2026). Wine online and digital decanting: Naturalness effects in large-scale consumer reviews of wine. Wine Economics and Policy. https://doi.org/10.36253/wep-19943

Open access under a Creative Commons CC BY 4.0 licence.

Related publications