Subscribe to DSC Newsletter

Data Science Meets Bubbly: What Data Says About Champagne Buying Patterns

Everyone loves champagne, right? But what strongly influences people’s behavior to purchase that bottle of bubbly? A growing body of research literature has found that a number of factors, including bottle color, label style and shape, and critics ratings all impact perceived value. By some measures, over 70% of wine buyers surveyed said those factors are “extremely important”, “very important” or “important” in their purchase decisions.

But as we all know, $5 Trader Joe’s organic wines, “Two Buck Chuck” and the Barefoot labels are popular so we weren’t so sure. After all, the idea that even the best critics can tell the difference between wines in a blind taste test has been thoroughly debunked. Ditto, the claim that price of wine matters in perceived quality.

So we built a little bubbly experiment. First, we used Import.io to extract data from over 300 different bottles of champagne and sparkling wine on Wine.com. This included critics ratings, images, and prices.

Screen Shot 2014-12-19 at 2.34.36 PM.png

Then we turned the data over to our CrowdFlower on demand contributors to look at each wine listing and answer some basic data enrichment questions: What color is the bottle? What shape is the label? What would you rate this wine on a scale of 1 to 100 based on appearance of bottle and label? Is the wine classy or not classy? And lastly, would you buy this wine? With these answers, the CrowdFlower contributors would be enriching our data and adding useful sentiment that we can analyze.

We then took the normalized results and mapped them back into a simple spreadsheet structure and put the data up into Silk.co. 

Explore the data here champagne-sentiments.silk.co

Our data science team concluded that a darker bottle color resulted in a 4% greater preference to buy a wine as compared to a clear / transparent bottle color. Oddly, white bottles had a 100% buy correlation (Anyone have any ideas on that? Please ping us). Similarly, we found that an etched bottle had a significantly diminished perceived sentiment. About 19% of respondents said they would not buy a wine with a label etched onto the bottle. Square labels got an 11% “No” rating. Round labels had the strongest positive correlation, with only 7.3% of respondents saying that they would not buy a wine with a round label. Not surprisingly, the perception of “Classiness” had the strongest influence on stated buying intentions. For bottles of wine that respondents had labeled as “No” to the classy question, 31% said they would not buy it. This is the highest correlation we saw although its fairly dubious - 69% of respondents would still have bought the wine!

You can play around with the above visualization to see our results or you can jump back into the data and see it directly or create other types of visualizations on our Champagne Sentiments Silk.

 

Views: 954

Comment

You need to be a member of Data Science Central to add comments!

Join Data Science Central

Comment by Sean Murphy on December 28, 2014 at 8:37am

If people said 'yes' to champagne bottles at a moderate rate in the population, it would be odd. But 86.8% of people said yes in the larger sample of dark bottles. If we took that as the average for the population, you'd have a ~37% chance of getting the 100% on a sample of 7 bottles, without indicating any difference between dark and white.

I'm curious, though, because it looks like more than one person rated each bottle, from the screenshot. If that's the case, why only a single Yes/No rating for each bottle in the silk?

Comment by Alex Salkever on December 27, 2014 at 11:16pm

Sean - good point. Still, 100% correlation is a bit odd even in that sample size, no? 

Comment by Sean Murphy on December 27, 2014 at 5:43pm

'Oddly, white bottles had a 100% buy correlation'

Well, there are only 7 white bottles in the dataset, so this is more than likely a small sample anomaly rather than a statement of the power of a white bottle. Interesting post though - I hadn't seen most of these tools before.

Comment by Alex Salkever on December 27, 2014 at 12:17pm

Thanks, Renette.

Here's the responsive embed of the data viz, as well. 

Videos

  • Add Videos
  • View All

© 2019   Data Science Central ®   Powered by

Badges  |  Report an Issue  |  Privacy Policy  |  Terms of Service