How Booking.com developed the first global vacation rental quality rating
While star ratings have long since set the bar in the hotel industry, the vacation rental segment has historically been somewhat fragmented in its ability to communicate quality to guests. But as the industry has grown, so too has the need for a widely recognized rating that would help guests make informed decisions and give properties an opportunity to improve their visibility on a global stage.
That’s exactly what we set out to deliver. We caught up with Inas Abdella, Director of Product for Home, to learn more about our vacation rental quality rating system and how it helps hosts get discovered quickly on our platform.
Why did Booking.com create a quality rating for the vacation rental industry?
When we looked at the experience of potential guests searching for an accommodation, we saw it was really difficult for them to discover vacation rentals based on a perceived quality level. They could look at other factors like photos, price, amenities, but there was nothing that differentiated one property from another in a similar way to what the star rating provides for hotels.
Guests had to spend time looking at different features to try and create in their own mind a model for what quality looked like, and it was very difficult for them to scan through, benchmark these observations, and compare different types of properties. Star rating was also one of the most used search filters among potential guests, which meant that a lot of vacation rental properties would simply not show up in the results. We wanted to better support guests in their decision-making process, help partners achieve more visibility for their properties, and set the right expectations to match them with the right guest.
This system is especially important for vacation rental partners who are new to our platform because it increases their visibility and helps potential guests find their property, even if they don’t have any guest reviews yet.
Previously, what was the alternative?
Some property management companies had their own rating systems applied to portfolios and labels that gave an indication of properties at one end of the spectrum. But there wasn’t anything globally recognized or set objectively by one party for the industry. We wanted to develop something so that no matter where the guest or property was based, there was a benchmark for comparison that would bring some consistency to how they could search for different accommodations. It was also key that this solution was scalable in nature, technology-based, and open to experimentation.
Why does an industry-wide classification matter?
Ultimately, a quality rating is more objective and focused on what the property offers. You can look at guest reviews for example, but that’s subjective and based on the individual’s personal experience – and in some instances, can be difficult for a host to influence. Quality ratings set a guest’s expectations, as opposed to reviews, which confirm whether the expectation was met. Focusing on quality and supporting hosts in identifying the things within their scope that might inform that rating will also help improve standards. We want to help our vacation rental partners understand why they receive a certain rating and the steps they can take if they want to improve it.
How does the quality rating work?
Setting a uniform global standard was a long process. Leveraging our global reach and relationships with partners, we explored a wealth of data, identified different patterns, and explored models to assess what we knew about our customers and their travel experiences. Through the application of modern machine learning technologies, we can understand what matters to guests. We looked at the characteristics that made vacation rental properties really stand out to inform the rating “recipe.”
We awarded quality ratings to 11 home and apartment-like properties (i.e. apartments, villas, vacation homes, guest houses, B&Bs, condo hotels, country houses, farm stays, chalets, riads, and gites) based on an algorithm that considers over 400 different features, including guests’ previous booking behavior, property attributes (e.g. facilities/amenities, property size, photos), and guest review score, if available. This algorithm also weighs various facilities and amenities differently depending on location – for example, a heater would be more important for a guest in Norway than in Bali. We also had to figure out how to continually validate and enrich that data in order to provide the best input for the ratings.
This is an ongoing process, and we continue to fine-tune it to further match guest expectations and adapt to market changes. It’s important that the rating is as accurate as possible, since overselling a property can lead to a negative guest experience. We recommend that partners enter info on their facilities, amenities, bathrooms, bedrooms, and sizes accurately and completely.
There were some interesting findings. For example, during development we made the assumption that properties with a higher rating had a higher cleanliness score. However, as we talked to guests and explored the data, we saw that cleanliness was a standard expectation regardless of the property’s quality rating. Even at the most basic level, cleanliness was a must. This became really important, so the properties that were assigned a rating have already passed certain checks to qualify – cleanliness being one of them.
How can partners influence their rating?
Properties with a quality rating can access personalized insights on the Extranet to be informed on why they received a certain rating and get advice to help them improve their quality. Generated by a machine learning model, these suggestions are a great place for partners to start if they’re looking to increase their appeal to potential guests and, in turn, improve their quality rating.
- Booking.com created a global vacation rental quality rating to set an objective standard for the industry that would support guests’ decision-making and help partners set the right expectations
- In addition to creating a consistent benchmark for comparison, the aim was to develop a rating that was scalable, technology-based, and open to experimentation
- The development process involved data deep dives, application of machine learning, and the creation of an algorithm that considers over 400 different features
- The quality rating is currently applicable to 11 home and apartment-like properties on our platform