May 6, 2020
Case Study

How to look good on video calls: analyzing 1K rated Skype & Zoom rooms. COVID-19 Voyeurism.

Victoriano Izquierdo
TLDR: Sorround yourself by 🌵plants , 🛋 lamps or a 🖼 piece of art if you don't want to look like you were taken 🔫 hostage. Cozy ornamental elements like 📚bookshelves, 🔥fireplaces, vintage kitchens are a big plus. Combine well the 🎨colors and tones of the walls with the rest of elements and your clothes. Natural light over artificial light. Make sure the 🤳angle and perspective of the camera don't make you look fat.

Just in a few weeks, 200K people started following the @RateMySkypeRoom or the Bookcase Credibility projects on Twitter. Every few minutes a new screenshot of someone on TV speaking from home gets analyzed. Claude Taylor, a veteran of presidential campaigns who worked for the Clinton White House, is behind it.

Since they have already rated almost 1000 Zoom and Skype rooms I thought that would be big enough to turn all these screenshots and reviews into an structured dataset to find some clear patterns on what makes people look good or bad on videocalls.

After getting all the tweets with our scraper Tractor, I uploaded the dataset to Graphext and filter out only the original tweets containing a rating between 0 and 10. Then I clustered all the ratings using our simmilarity algorithm for short texts, which uses word2vec.

The main narratives

So each cluster represents here a "narrative". Ratings that share a semantic (the camera angle, cozy elements, the colors and tone...) get connected. High dense regions of the networks are delimited as clusters. We define the clusters with the terms that appear more often in that cluster and are less generic in the entire corpus.

Here you have a detailed definition of each cluster using the most significant 📚🖼 nouns and 👍👎adjectives. To see it bigger click here

What defines the best and worst ratings?

What are the most popular ratings?

Although the big mayority of ratings are positive, between 6 and 9, guess what... the most popular reviews are the ones with the highest score (10) and the lowest (0,1). Extremes are exciting, normality is boring. But if all the ratings were extremes they would become normal...

Here you have some examples of some of the most retweeted ratings

Go and play yourself with the data in this Graphext project (works much better on desktop)

And here is big sample of picture including the number of Favs and RTs each one of them got. Click here to see it bigger.

Maybe we are all "pretending" these days, but at least let's pretend well!

View image on Twitter
The New Yorker


A digest of our blog data analysis, product updates and company news
Thank you! Your submission has been received!

Sorry. Something failed

Other stories