Hosting in Auld Reekie
Visualizing the AirBNB market in Edinburgh, Scotland.
Since its inception 15 years ago, Airbnb has exploded from a single rental to over 5.6 million listings and 150 million users (Meyer, 2022). It has changed the way people travel and find accommodations (myself included), and has become one of the most prominent examples of the growing share economy (Talcum, 2019). Having used Airbnb many times as a guest, I wanted to take a look at the flip side of the Airbnb experience for this project: hosting on Airbnb.
Context
From what I’ve been able to gather without actually being a host myself, Airbnb provides a performance dashboard that tracks metrics such as average occupancy rate, nightly price, listing conversions, and guest ratings (Rusteen, 2020, 0:10:04). Hosts can also see the target metrics for superhost status and how they’re doing in each of those categories.
The dashboard includes a simple line graph which shows how the host’s performance has changed over time, and how they measure up against other “similar listings.” However, Airbnb doesn’t explain how they define similar listings apart from saying that they are listings “within the same geographic location” (Rusteen, 2020, 0:02:34). So hosts aren’t able to dig any further into this data to find out who they’re being compared to or why their performance might be different than other hosts’ performance.
I wanted to create a dashboard that would allow hosts to explore larger trends in the Airbnb market in their region and filter down to the individual listing level so they could see what other hosts might be doing differently.
Data
I used data from a site called InsideAirbnb which provides detailed listing information by city. (Data can be accessed here.) I decided to focus on the Airbnb market in Edinburgh, Scotland for this project since it’s a city where I’ve had great Airbnb experiences in the past. The dataset I selected contained over 7,500 Airbnb listings with details about price, location, room type, the number of people accommodated, ratings, superhost status, and many other attributes.
I cleaned up the data in Excel, eliminating any listings with incomplete information. I also excluded several extreme outliers that didn’t make sense (e.g. $8,425 per night for one bedroom with a shared bathroom).
I also added several derived attributes. The dataset provided the nightly cost for each listing and the number of people that the listing accommodates which I used to calculate the price per person. This makes it easier for hosts to compare pricing across the board. I also looked at the number of listings per host and created three categories: conventional host (1–5 listings), semi-professional (6–10), and professional (more than 10 listings). Professional hosts with tens or hundreds of listings might use different strategies than conventional hosts, so I wanted users to have the ability to filter the data accordingly.
Literature Review
After cleaning up and getting familiar with the data, I explored the work that other people had conducted around the topic of Airbnb visualizations and on hosting in particular. During the review, price repeatedly came up as one of the biggest factors which hosts can use to influence their bookings (Lladós-Masllorens et al., 2020; Setiawan & Diani, 2021).
Many of the other projects I explored were aimed at helping guests find the perfect Airbnb listing to book. While their target audience was different, these projects provided many interesting visualization strategies which I thought could be useful for hosts as well. The most common visualizations included maps (Yichong, 2019; Oktay, 2022) and scatterplots (Gupta, 2019). I knew that I wanted to create visualizations that would be familiar and easy for hosts to read, so exploring visualizations that were created for the general population (potential Airbnb guests) was valuable.
The literature review also helped me narrow down the level of detail I wanted to include in my visualizations. Some of the visualizations I explored used summarized data to make it easier to gather insights about higher level Airbnb trends, but I decided to focus on visualizing individual listings rather than averages or summaries so that hosts would be able to see the details of each listing and compare it to their own.
Ideation
In the sketching phase, I began thinking about the types of visualizations that would be most valuable for Airbnb hosts. I sketched out a variety of visualizations that conveyed data ranging from price and location to booking information and guest reviews.
I narrowed down my ideas by focusing on data attributes that hosts could most easily control. I wanted the information they gathered from the dashboard to be actionable for them, so I decided to focus especially on price per person, superhost status, and amenities offered.
Evaluation and Iteration
From there I brought my data into Tableau and started developing three visualizations: a scatterplot matrix to examine the correlations between price and various other factors, a map to explore price by location, and a bar chart to display amenities provided. I evaluated my dashboard with three participants and used their feedback to make improvements to the visualizations.
The matrix layout of the scatterplot visualization was a common source of confusion for all three participants. There was a lot going on in a fairly small space and the participants were unsure of which parts of the matrix were relevant. As a result of this feedback, I reduced the scatterplot matrix to two side by side scatterplots which examined how price per person correlated to the number of guests accommodated and value ratings.
Another reoccurring source of confusion was the color coding of the marks on the map. I was using color to indicate which neighborhood each listing was located in, but participants assumed that the color had something to do with price. To make the visualization more intuitive for viewers, I decided to use a color gradation to indicate price per person.
Final Dashboard
The final result is a dashboard which helps hosts explore how they fit into the larger Airbnb trends in Edinburgh. It also provides detailed information so that they can drill down to find out what other hosts are doing and then use that information to intentionally position themselves within the market.
Final Release: Interactive Tableau dashboard
Hosts can use the scatterplot and map visualizations to get a feel for the range of prices that hosts charge per person. The scatterplots and bar chart are also color coded to indicate superhost status (orange for superhosts and blue for standard hosts). I was not able to implement the same color coding on the map because of limitations in Tableau. Ideally I would have been able to use a blue gradation to indicate the price range for standard hosts, and an orange gradation to show the price range for superhosts, but I was not able to separate out the color and saturation channels so the blue gradation indicates pricing for all listings.
Clicking on a listing in the scatterplot will highlight that listing on the map (and vice versa) which allows hosts to explore the same listings from multiple perspectives. Hovering over an individual listing brings up a tooltip which includes listing details as well as a URL so that hosts can navigate to the actual listing on the Airbnb site and view additional details such as photos and listing descriptions.
The bar chart at the bottom shows a sampling of different amenities that hosts offer. Hosts can use this bar chart to make sure that they provide common amenities such as a washer and wifi that guests are likely to expect. They can also look at the less common amenities, for example a dedicated workspace and luggage drop off, for ideas of amenities they could offer to add value and make themselves stand out from their competition.
The dashboard also includes what is perhaps the key component for providing actionable insights for hosts: a collection of filters which allow hosts to narrow down the data and explore listings similar to theirs. Hosts can filter the data by neighborhood, number of people accommodated, price, superhost status, room type, as well as several other attributes. This allows them to explore their competition on a more granular level and then use that information to make decisions about their own listing.
- Gupta, S. (2019, January 4). Airbnb Rental Listings Dataset Mining. Medium. https://towardsdatascience.com/airbnb-rental-listings-dataset-mining-f972ed08ddec
- Lladós-Masllorens, J., Meseguer-Artola, A., & Rodríguez-Ardura, I. (2020). Understanding peer-to-peer, two-sided digital marketplaces: pricing lessons from Airbnb in Barcelona. Sustainability, 12(3). https://doi.org/10.3390/su12135229
- Meyer, S. (2022). Airbnb statistics and host insights [2022]. The Zebra. https://www.thezebra.com/resources/home/airbnb-statistics/
- Oktay, E. (2022, January 17). Step-by-Step: Visualizing Airbnb data of New York. The Information Lab. https://theinformationlab.nl/2022/01/17/step-by-step-visualizing-airbnb-data-of-new-york/
- Rusteen, D. V. (2020, June 3). _ The New Airbnb Performance Dashboard: Learn What Makes It Useful [Video]. YouTube. https://www.youtube.com/watch?v=F-2ANCNh8HI&list=WL&index=2
- Setiawan, I., & Diani, F. (2021). Visualization of Amsterdam Airbnb Business Performance using Customer Reviews. International Journal of Applied Sciences in Tourism and Events, 5(2), 142–152. https://doi.org/10.31940/ijaste.v5i2.142-152
- Talcum, S., Jr (2019). The Sharing Economy Is Still Growing, And Businesses Should Take Note. Forbes. https://www.forbes.com/sites/forbeslacouncil/2019/03/04/the-sharing-economy-is-still-growing-and-businesses-should-take-note/?sh=690e27d64c33
- Yichong (2019, December 12). Airbnb in D.C. and Beijing. Medium. https://medium.com/visumd/what-i-wish-id-known-when-considering-short-term-rentals-e83641e46ca4
