Thread by @kate_ptrv

I couldn’t just walk past this Tweet, so here is some fun #dataviz

Scented candles: An unexpected victim of the COVID-19 pandemic 1/n

First, I downloaded a random subsample of US-based customer reviews of the 3 most popular scented candles on Amazon. Between January 2017 and January 2020, the average rating stayed around 4.3/5, but there was a sharp drop between January and November 2020. 2/n

Could it be because of the COVID-related loss of smell? To investigate, I plotted the reviews of the 3 most popular unscented candles on Amazon*, and the difference was quite striking 3/n

*One thing to note: fewer reviews are available for unscented candles than for scented ones

Since the beginning of 2020, customer satisfaction with scented candles has been dropping at a much faster rate compared to unscented candles. 4/n

That’s where things get really interesting. Instead of just looking at ratings, I decided to dig into the content of these reviews, so I wrote a quick function that flagged all reviews that mentioned lack of scent. 5/n

Now, most people were rating their candles without taking the time to write a detailed review, so I expanded my sample to include top 5 scented candles from 4 different brands. 6/n

Since the beginning of this year, the proportion of reviews mentioning lack of scent grew from < 2% in January to close to 6% in November. These numbers may not seem like much, but the trajectory of no-scent reviews over the last 11 months is certainly an interesting one 7/n

So, if you are doing any virtual Black Friday shopping today, and if you still have your sense of smell, maybe buy a scented candle and leave a nice review. And no, I don't work for a candle company. I just like looking at data. 8/n

Data + code examples: github.com/kateptrv/Candl…

Chrome plugin to download your own Amazon reviews data: chrome.google.com/webstore/detai…