Why Airbnb Made Such a Big Deal About Categories

From Search based to Discovery first

Last week Brian Chesky announced “the biggest change to Airbnb in a decade” with “Airbnb Categories”. At first glance, Categories don’t seem like an innovation worthy of such a strong statement, so what was Brian thinking? Behind this product feature, the core of the announcement points to a deeper problem that affects companies today.

I don’t search for what I don’t know exists”

Search works well when a user knows what they are looking for and can articulate it in a query that makes sense to a computer — but often they can’t do either. For information or Q&A based products like Google Search, typing the query “How to build a treehouse” can be meaningfully answered on the first result. The answer on page 99 is likely to be genuinely less relevant. However, for products that are commerce or entertainment based the user-experience of search diminishes as the amount of relevant options exceeds hundreds or thousands of possibilities. The unknown content in unknown categories cause discovery issues for users that leave them limited to what they already know exists.

For a commerce based product like Airbnb, hidden in the long tail of accommodation listings are exciting and inspiring options that need discovery-based product experiences to create new mental models for users of what’s possible on the platform. Without knowing that ‘Camping’, ‘Arctic’ and ‘OMG!’ are options many people, including myself, are highly unlikely to query for them in a search term. After a long hiatus, Categories has made me genuinely excited to use Airbnb again to book holidays, team off-sites and inspire me to travel. This is precisely the power of discovery.

Discovery-based experiences like Airbnb Categories are those which companies attempt to predict what a user wants to engage with without an explicit search query.

Discovery is the new king - Tiktok is now more popular than Google Search

It seems like an impossible feat but TikTok with its “For You Page” overtook Google Search as the most popular online destination in the world in late 2021. Odds are if you’re reading this you already know that when you open TikTok, which videos are shown to you are determined by 1000s of data points such as what you’ve watched previously, who you follow, the category of video and what city you are in. In order to offer these types of discovery-based experiences, large tech companies have built complex machine-learning infrastructure to rank content that is most meaningful to each user.

While Airbnb had already implemented machine-learning and ranking systems in their search experience, Brian’s announcement tells us that they are adapting their product to the global trend of discovery first experiences.

Discovery first means products must adapt quickly to user interactions and behavior

While it’s commonplace for companies to collect and track millions of data points for analytics, few are able to use the data to immediately improve their product. This is because analytics is accompanied with a substantial amount of inertia. Human synthesis, insight and manual implementation are required to act upon it. That’s not easy.

With analytics, the typical cycle for a company to understand what user behavior is occurring is daily, weekly or monthly at best. Implementing new product features is on an even longer timescale from monthly, quarterly or yearly. In contrast, discovery first products are set up in a way that they are always improving with machine-learning. Companies like Tiktok, Netflix, Youtube, Amazon and Facebook have set up incredible ranking systems with feedback loops that adapt in seconds, minutes and hours to user input and behavior. For example, the Explore tab on Instagram makes 90 million model predictions every second. Discovery first products that adapt quickly to user interactions and behavior hold an enormous advantage over products that solely offer search or manually curate content. In order to engage, grow and provide users with relevant content, every company needs machine-learning infrastructure.

Discovery is changing the world

With an unimaginable amount of content on the internet, machine-learning infrastructure that improves discovery and relevance have changed the way products are built. Companies like TikTok implemented these systems from inception to improve activation, conversion, engagement and retention.

Discovery first is happening everywhere across industries and use cases from marketplaces, media and entertainment, social media, e-commerce and education. The machine-learning infrastructure that powers these experiences is having a significant impact on business metrics as well as inspiring, entertaining and delighting users.

Airbnb is just the latest example of this trend towards discovery first changing the world. As Brian said, Categories represents “a new way to search that makes it easy to discover millions of homes you never knew existed.” Thanks to this, I’m feeling inspired to book some exciting travel this year that I otherwise wouldn’t have even known was possible.

Get up and running with one engineer in one sprint

Guaranteed lift within your first 30 days or your money back

100M+
Users and items
1000+
Queries per second
1B+
Requests

Related Posts

Nina Shenker-Tauris
 | 
September 5, 2024

Is the key to unlocking better user experiences in recommender systems found in exploration?

Tullie Murrell
 | 
March 1, 2023

Evaluating recommendation systems (mAP, MMR, NDCG)

Jaime Ferrando Huertas
 | 
September 19, 2022

Day 1 of #Recsys2022: Our favorite 5 papers and talks