Ai Editorial: What’s the secret behind Airbnb’s end-to-end travel platform?

First Published on 12^th March, 2019

Ai Editorial: Rather than being an avenue only for accommodation, Airbnb is evolving to be an end-to-end travel platform. Ai’s Ritesh Gupta understands what is being done to power all of its verticals.  

In the second half of 2018, Airbnb made a couple of big announcements – crossing the 400 million-mark in terms of its total guest arrivals and 3.5 million guest arrivals in a single night in August. Earlier this month, the company stated that it has over 6 million homes. Airbnb is expecting half a billion guest arrivals in Airbnb listings by the end of first quarter.

A question that is being probed is - how Airbnb is gearing up to take the so-called network effect to the next level as it goes deeper into the booking funnel.

As it develops into an end-to-end travel platform which essentially means focusing on where to stay, what to do, and how to get there, what this would mean for the traveller?

Airbnb is definitely one of those brands that the entire fraternity believes can provide a better answer to the fragmented travel shopping experience. The way Airbnb goes about uniting its staff/ teams across product disciplines, making the most of “creative hacks” or managing complexities associated with product designing, it can provide an answer to when to travel, where to go, and what to do on trips.

Thinking like a traveller through “entities and relationships”

The best part about Airbnb’s platform is to let travellers get closer to taking a decision.

This means that Airbnb is coming up with the best possible result for users on its platform when it understands their intent. This is going to be based on setting up an environment for accessing structured data and enabling users to enjoy their trip planning and buying by a gamut of choice being presented to them. In order to attain this, the expertise of engineering and data science comes into play. The value of context i. e. to ensure content served or inventory presented clicks with the user.

For this to work, as explained by Airbnb, it is imperative to work on a meticulous way of ascertaining relationships between discrete but related entities such as cities, activities, cuisines, etc. Queries of users would be answered aptly or precisely by storing details about entities and the association between them.

To work this out, Airbnb has been refining its knowledge graph to classify its inventory and contextualize the entire platform. The focus here is on storing and serving structured data that connects what makes Airbnb’s inventory distinctive, what travellers are searching for, and what is there to be explored or availed. The concept of knowledge graph isn’t new. For instance, Google’s knowledge graph intends to offer precise information in response to users’ search queries, within the shortest possible time.

Generally relational databases store data in tables. Tables can shape up to have massive details, and have an array of columns and records. Pointing an issue with this, Airbnb mentioned that  there is an “operational burden when you have many table for distinct objects that may contain the same relational information in individual columns”. Referring to the utility of the graph structure, the Airbnb team says structuring queries in terms of this graph paves way for optimal data semantics maintenance. Citing an example, Airbnb referred to surfing – as an experience it should be the same surfing that a destination like Hawaii is known for. Such structuring around the relationships between the entities provides the scalability and flexibility needed to expand categorization to any number of things, mentioned the team associated with this project. This helps in avoiding duplicate data. The taxonomy in the knowledge graph has been applied it to classify all of available inventories at Airbnb.

One of the highlights of the graph is presence of nodes (could be entities such as restaurants, experiences etc.) and edges (the sort of relationships that exist between the entities). There are various types of nodes for diverse set of entities and different types of edges for different types of relationships (located in, tagged by, etc.). The team also acknowledges that an initiative like tagging restaurant with the relevant nodes isn’t easy.

“From there, we have a flexible API to query for neighbors connected by certain types of relationships and can index our inventory items by the unique identifiers of their corresponding representation in the knowledge graph,” according to Airbnb.

Airbnb asserts that this is a solid foundation to understand what one’s trip possible could be all about. Be it for designing of the interface, how the options are going to be presented to actually working on contextual travel insights to users in the booking flow is a fascinating journey. The action that a user can take could be finalizing a location to travel, a home to book etc. It would be worth watch how Airbnb shapes up this foundation further to add more trip essentials and excels in its quest of becoming an end-to-end travel platform.  

Hear from senior industry executives about how the industry is looking at offering the “end-to-end journey” solution at this year’s Ancillary Merchandising Conference, scheduled to take place in London, UK (9-11 April, 2019).

For more info about Ancillary Merchandising Conference, click here

Follow Ai on Twitter: @Ai_Connects_Us