One way Airbnb helps hosts succeed is by observing search and booking trends and offering insights and tips based on those observations. Deepank Gupta, an engineering manager, and Kidai Kwon, a software engineer, recently shared just how they gather and share this information. Here’s a sneak-peek into the process!
Why we deliver insights
We want to make sure the insights you get boost your hosting success, so we decided that to be share-worthy, each insight has to be:
- Personalized: Using machine learning algorithms, each insight takes into account the characteristics of your listing and market, such as current demand and booking probability for this type of space and location.
- Targeted: You’ll only receive a particular insight if your listing matches certain criteria such as listing attributes and pricing settings, so every insight you see should be relevant to you—and we’re thoughtful about how many times you see this information.
- Actionable: You should be able to take immediate and specific action based on each insight we share.
For example, one of the most common questions we hear from hosts is: How do I pick the right price? Setting a price can be hard without reliable information about other listings in your area, travel trends, and the interest people have in the amenities you offer. In the past couple years, we launched pricing tips and Smart Pricing that help hosts get personalized, daily pricing tips right in the calendar. While the price you set is ultimately always up to you, and you can override Smart Pricing for any night in your calendar at any time, we think it’s a helpful tool that’s being used by many hosts.
To do this, we developed a mathematical model that learns how likely a guest is to book a specific listing, on specific dates, at a range of different prices. It uses different types of information, including listing type, location, current price, availability, and how far off each available date is. As our models have gotten better at understanding the market, we’ve been able to pass along the learnings to you. For instance, if travelers tend to look for long term stays in an area, we can recommend setting discounts for longer bookings.
How we share the insights
In order to deliver insights to hosts’ dashboards, calendars, and inboxes, we built a behind-the-scenes, or “backend,” system that collects publicly available information from around the web. We affectionately call this system Narad—the name of a divine Indian sage and traveling storyteller who carried news and enlightening wisdom. Our own Narad system is designed to generate personalized insights, to rank their effectiveness by listing type and context, and to deliver those insights to you at the right place and right time.
Here’s how it works:
1. When a request for insights is received, the backend system collects all the insights applicable to the listing.
2. Narad filters out irrelevant insights based on host and listing information, and on prediction models.
3. Next, it filters the insights based on what we’ve shown you before, so you see fresh and relevant information.
4. For each insight Narad thinks will be useful, it generates personalized information such as potential booking gains or a tip for discounts on longer stays based on data from our machine learning algorithms.
5. Finally, the insights are ranked on how valuable they are to hosts. This way you’re most likely to see the insights with the most potential impact for you.
Another way to listen to hosts
When we pay attention to what matters to you, we learn how to deliver the recommendations you want. While we’ll always gather feedback the old-fashioned way, through discussion and engaging with the ideas you share personally, we can also learn from the choices you make. One way the Narad system helps us listen to you is by tracking which insights you take action on. The system takes note of the things you click, dismiss, and investigate. Over time we learn how to offer you an enjoyable experience and relevant content.
So far, the Narad system has made a positive impact on host success. For example, we’ve seen an overall lift of 2% in bookings attributable to implementing insights on the host calendar page. And 1 out of 5 hosts have acted on insights we’ve offered, so it’s clear there’s an appetite for more. To make sure you get the information that benefits you most, here’s how we plan to expand this work:
- Increase personalization and capture more macro market trends to deliver a broader range of insights.
- Create better ranking algorithms so you always get insights with the most relevance and impact.
- Bring insights to more host tools and platforms.
Your success is our priority, and we’ll continue to build with hosts in mind. We hope you’ll continue to let us know how our insights are working for you.