This article illustrates how we at Angel Host use advanced analytics to optimize the properties we manage. Data is at the heart of our decision-making processes, and this project demonstrates the value of such data — and the expert interpretation of it — to deliver the best results for our clients around the world. What follows is a summary of one of our listing optimization projects completed at the end of 2022.
The project focused on understanding the answer to a critical question:
What do properties that appear high on Airbnb’s search results have in common?
More specifically, we aimed to uncover the correlation between a property’s attributes and its ranking on the search results page. While we know, based on academic papers published by Airbnb, that the most important elements to obtain a high rank are relevancy and conversion, we wanted to understand what elements likely lead to a higher conversion.
In other words, by understanding the characteristics of the top-performing properties, we could then ensure our listings had those characteristics, thus improving our performance as hosts and ultimately allowing us to provide better returns for our clients.
To answer these questions, we needed to gather large amounts of data to form the basis of our research. This involved aggregating the results of hundreds of searches and gathering over 70 unique variables for each of the thousands of properties that were analyzed.
These variables ranged from the number of photos, amenities, reviews, and nightly prices, to more specific variables such as the age of the listing, the number of words used in the description, whether the listing had a carbon monoxide detector and even the number of landscape photos the listing had!
Our research project revealed something we had anticipated but have now definitively proved:
The correlations between property attributes and search rank varied by geographical location.
In other words, the attributes of top-performing properties vary by location. While many attributes are important no matter what market the property is in (e.g. having the right price), there are others such as the number of ratings that are very important in some markets and not so much in others.
Instead of using a top-down approach to search rank, a more organic ground-up strategy is employed by Airbnb, where the search results were unique to the market and likely based on the attributes that performed best in previous searches for every particular type of guest.
The knowledge of which attributes are most important in each market gives Angel Host a crucial competitive advantage, as we now optimize around these variables differently depending on what is most correlated to search rankings in each market. The chart below highlights how differently the variables correlated to search rankings depending on the geographical location.
The data highlights how different variables have varying correlations depending on the city. For example, the base_price variable is very highly correlated to search rank for city 1 and 2. Despite being an important variable for city 3, it is not the most highly correlated variable (see the points to the right of the chart). On the other hand, for City 3, we can see that having positive reviews far outweighs in importance to having the best possible price, which we have seen happening in cities with large quantities of property inventory but very inconsistent levels of quality.
Similarly, the active_listing_count (the number of listings a host manages) for city 2 is very highly correlated with search rank, however, in city 1 it is far less so. This doesn’t mean that Airbnb prioritizes in the search results hosts with multiple listings, instead, it likely means that in City 2, hosts with multiple listings (likely professional Property Managers) are preferred by guests and thus attract a larger portion of bookings.
There is of course overlap across the cities and in general, the trends are similar. Price and rating scores will always influence booking and therefore search rankings. However, there were some surprising differences across the cities. In some cities, the presence of large discounts for long-term stays was highly correlated to better rankings, whereas in other markets the presence of a discount was not correlated to search rank and was in fact negatively correlated!
Other interesting results included the fact that Airbnb seems to give new listings on their platform a rankings boost. The idea is that new properties don’t have any reviews or positive feedback to rank higher in the traditional ways, therefore they are given an opportunity to perform when they are new to the platform. This also serves to encourage new hosts to stick with Airbnb once they have proven that their platform works for the host too.
Finally, the results also showed that the presence of the “Rare find” or “New lower price” banner correlated with better search rankings. For the “Rare find”, there is an element of a self-fulfilling prophecy, as those properties that are rare finds will likely remain booked if they continue to be featured at the top of the search rank page. The only way to obtain one of these banners is to ensure the calendar is often booked, something that can only be obtained with the right pricing strategy. Similarly, obtaining a “New lower price” banner can be powerful, but not when done at the expense of profitability. Again, something that a good pricing strategy can help influence.
In conclusion, while top ranking is based on relevancy and conversion rates, high conversion rates are achieved by ensuring that property listings follow the best practices for each particular market.
If you want your listings to always outperform your market, these are some of the most important things for you to do:
If you’re curious to learn more about this study, feel free to reach out to us.